Substation point location information checking method and system based on AI identification fusion

Abstract

The invention discloses a substation point location information checking method and system based on AI identification fusion, and relates to the technical field of artificial intelligence. The method comprises the following steps: after hardware connection check in substation point location information check is completed, selecting a first point location to carry out telemetering test and remote control test to obtain first telemetering precision and first remote control precision; according to the moving window, selecting a point location electrically connected with the first point location, performing iterative telemetering test and remote control test, and constructing a telemetering precision matrix and a remote control precision matrix; calling historical point location check data of point locations of the same family, and performing precision check on the telemetering precision matrix and the remote control precision matrix to obtain a test credibility matrix; according to the test credibility matrix, the telemetering precision matrix and the remote control precision matrix, AI fusion identification is carried out, and a point location checking quality parameter matrix is obtained and serves as a substation point location information checking result. According to the invention, the precision and efficiency of substation point location information checking are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, in particular to a substation site information checking method and system based on AI recognition fusion. BACKGROUND

[0002] With the upgrading of the power system to intelligence and automation, the operation stability of the substation, as the core hub of power transmission, directly determines the safety of power supply. After the substation is put into use, testing and checking the remote control and telemetry functions of the internal electrical connection structure and information point become a key operation and maintenance link to ensure the operation quality. At present, the checking method in the industry mainly relies on direct testing, that is, by carrying out telemetry and remote control test to collect parameters of a single point to complete the basic information checking work.

[0003] However, the existing checking technology only focuses on the direct test parameters of a single point, does not consider the correlation of the electrical connection between points, and does not check the reliability of the test data. The single mode is easily affected by factors such as equipment instantaneous fluctuation and environmental interference, resulting in accidental errors, which in turn leads to deviation of the checking result, and cannot accurately reflect the real operation state of the substation point, thereby affecting the accuracy and efficiency of the substation point information checking, and even may cause safety hazards for subsequent operation and maintenance. SUMMARY

[0004] The present application provides a substation site information checking method and system based on AI recognition fusion, aiming to solve the technical problem of insufficient accuracy and efficiency of substation site information checking in the prior art.

[0005] In view of the above problems, the present application provides a substation site information checking method and system based on AI recognition fusion.

[0006] In the first aspect, the present application provides a substation site information checking method based on AI recognition fusion, comprising: After the hardware connection checking in the substation site information checking is completed, a first point is selected for telemetry test and remote control test to obtain a first telemetry accuracy and a first remote control accuracy; According to a moving window, a point connected to the first point is selected for iterative telemetry test and remote control test, and a telemetry accuracy matrix and a remote control accuracy matrix are constructed; The historical point checking data of the same family point are called to check the accuracy of the telemetry accuracy matrix and the remote control accuracy matrix, and a test reliability matrix is obtained; According to the test reliability matrix, the telemetry accuracy matrix and the remote control accuracy matrix, AI fusion recognition is carried out to obtain a point checking quality parameter matrix as the substation site information checking result.

[0007] In a second aspect, the present application provides a substation site information checking system based on AI recognition fusion, comprising: A first point precision test module is configured to, after the hardware connection checking in the substation site information checking is completed, select a first point for remote measurement test and remote control test, and obtain a first remote measurement precision and a first remote control precision; An iterative test module is configured to, according to a moving window, select points electrically connected to the first point, perform iterative remote measurement test and remote control test, and construct a remote measurement precision matrix and a remote control precision matrix; A precision verification module is configured to call historical point checking data of points of the same family, perform precision verification on the remote measurement precision matrix and the remote control precision matrix, and obtain a test credibility matrix; An AI fusion recognition module is configured to, according to the test credibility matrix, the remote measurement precision matrix and the remote control precision matrix, perform AI fusion recognition, and obtain a point checking quality parameter matrix as a substation site information checking result.

[0008] One or more technical solutions provided in the present application have at least the following technical effects or advantages: The present application provides a substation site information checking method and system based on AI recognition fusion. The basic precision data is obtained by first performing remote measurement and remote control test on a first point, then the associated points are iteratively tested by a moving window to construct a double precision matrix, then the matrix credibility is verified by calling historical data of points of the same family, and finally a quality parameter matrix is generated by AI fusion of multiple matrix data. This method not only avoids accidental errors of single test, but also realizes comprehensive correlation and accurate verification of data. The present application enhances the comprehensiveness and accuracy of point state evaluation, improves the automation level and processing efficiency of checking, and thus effectively improves the precision and efficiency of substation site information checking. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0010] Figure 1 A flowchart of the substation site information checking method based on AI recognition fusion provided by the embodiments of the present application is shown. Figure 2 A structure diagram of the substation site information checking system based on AI recognition fusion provided by the embodiments of the present application is shown. In the drawings, the components represented by the numbers are described as follows: The first point precision test module 11, the iteration test module 12, the precision verification module 13, and the AI fusion identification module 14. DETAILED DESCRIPTION

[0011] The application provides a substation point information checking method and system based on AI identification fusion, and aims to solve the technical problems of insufficient precision and efficiency of substation point information checking in the prior art.

[0012] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the application.

[0013] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server comprising a series of steps or units does not have to be limited to only those steps or units clearly listed, but can include other steps or modules not clearly listed or inherent to the process, method, product or device.

[0014] Embodiment one, as shown in the application provides a substation point information checking method based on AI identification fusion, which comprises: Figure 1 S100: after the hardware connection checking in the substation point information checking is completed, a first point is selected for remote measurement test and remote control test to obtain a first remote measurement precision and a first remote control precision. S100: after the hardware connection checking in the substation point information checking is completed, a first point is selected for remote measurement test and remote control test to obtain a first remote measurement precision and a first remote control precision.

[0015] In the embodiments of the application, after the hardware connection checking in the substation point information checking is completed, a first point is selected for remote measurement test and remote control test to obtain a first remote measurement precision and a first remote control precision. The core of the substation point information checking is to verify the matching of the correctness of the hardware connection and the accuracy of the information function. The hardware connection is the physical basis for the information point to realize the remote measurement and remote control function. If the hardware connection does not match the design drawing, the subsequent remote measurement and remote control test will lose its meaning, and even the equipment may be damaged due to the wrong connection. Therefore, the hardware connection checking must be completed and confirmed to be correct before the function test of the information point is carried out. The first point is selected as the initial test point to establish the reference precision data, provide a reference for the iteration test of the subsequent associated points, and avoid the test confusion caused by the lack of reference.

[0016] The step S100 in the method provided by the embodiments of the application comprises: After the hardware connection check in the substation site information check is completed and the check result is correct, a first point is selected for remote measurement and remote control by reading the substation design file; According to the test standard of the electrical element corresponding to the first point, remote measurement test and remote control test are performed to obtain a first remote measurement accuracy and a first remote control accuracy.

[0017] First, after the hardware connection check in the substation site information check is completed and the check result is correct, a first point is selected for remote measurement and remote control by reading the substation design file. The hardware connection check refers to verifying the consistency of the physical connection of the electrical element with the design file, which is the premise of information-based testing. The electrical element includes cables, terminal blocks, transformers, etc. The physical connection includes cable model, terminal number, port correspondence, etc. The first point refers to the reference point for initial testing, usually the information-based point corresponding to the core functional element, which is convenient for subsequent correlation testing. The core functional elements include transformers, circuit breakers, etc. After completing the hardware connection check of the electrical elements in the substation and confirming that the connection relationship is completely consistent with the design drawing and there is no wrong connection or missing connection, the substation design file is read to select representative core information-based points as the first point for remote measurement and remote control testing. For example, the secondary cable of the current transformer TA1 in the 10kV outgoing line cabinet of a certain 110kV substation is checked, and it is confirmed that it is connected to the terminal AI1 of the measurement and control device and the secondary wiring Figure 1 is correct, the hardware check is passed. The point code table is read to determine that the information-based point code corresponding to TA1 is P101, the function is to remotely monitor the current and control the protection setting value to be put into or taken out, and P101 is set as the first point.

[0018] Secondly, according to the test standard of the electrical element corresponding to the first point, remote measurement test and remote control test are performed to obtain a first remote measurement accuracy and a first remote control accuracy.

[0019] Among them, according to the test standard of the electrical element corresponding to the first point, remote measurement test and remote control test are performed to obtain a first remote measurement accuracy and a first remote control accuracy, including: According to the test standard of the electrical element corresponding to the first point, a preset variable is set for the first point, remote measurement test is performed to obtain a first remote measurement variable, and remote control of the first point is performed with a preset control instruction to obtain a first remote control result; The similarity of the first remote measurement variable and the preset variable is calculated as the first remote measurement accuracy; The similarity of the first remote control result and the preset control result in the preset control instruction is calculated to obtain the first remote control accuracy.

[0020] First, according to the first point corresponding to the electrical element test standard, the first point is set to a preset variable, and the remote measurement test is carried out to obtain the first remote measurement variable, and the preset control instruction is remotely controlled to obtain the first remote control result. The electrical element test standard refers to the file that specifies the test parameter range and operation specification. The preset variable refers to the reference parameter value of the remote measurement test, such as current, voltage, which is used to judge the accuracy of collection. The preset control instruction is a standard control instruction for the function of the point. The first remote measurement variable refers to the parameter value actually collected by the measurement and control system. The first remote control result refers to the actual state of the element after responding to the instruction. According to the industry test standard of the electrical element corresponding to the first point, the preset variable conforming to the element operating range is set to the point, the remote measurement test is carried out through the measurement and control system, and the first remote measurement variable actually collected is recorded. At the same time, the preset control instruction is sent to the point, the element response state is observed, and the first remote control result is recorded.

[0021] For example, according to the electrical element test standard, the preset variable 400A analog current is injected into the P101 point, and the actual current value 398A is collected after remote measurement test, that is, the first remote measurement variable is 398A. The preset control instruction is sent to P101: TA1 overcurrent protection setting value is completed within 10 seconds, the measurement and control system displays "successful application"; the protection device is actually observed to complete application in 12 seconds, the measurement and control system completely displays "successful application", and the first remote control result is recorded as: 12 seconds response, and the state display is "successful application".

[0022] Secondly, the similarity of the first remote measurement variable and the preset variable is calculated as the first remote measurement accuracy. The remote measurement accuracy is an index for quantifying the accuracy of remote measurement function, and the value range is 0-100%, the higher the value, the more accurate the remote measurement. The similarity of the first remote measurement variable and the preset variable is calculated by using a quantitative formula, and the similarity is directly used as the first remote measurement accuracy, and the formula is: remote measurement accuracy=(1-|first remote measurement variable-preset variable| / preset variable)×100%. For example, the preset variable is 400A, the first remote measurement variable is 398A, and the formula is: remote measurement accuracy=(1-|398-400| / 400)×100%= (1-2 / 400)×100%=99.5%, that is, the first remote measurement accuracy is 99.5%. The deviation of the remote measurement data is converted into an intuitive accuracy index, avoiding fuzzy judgment, and providing a quantitative conclusion for the accuracy of the remote measurement function of the first point.

[0023] Finally, the similarity of the first remote control result and a preset control result in the preset control instruction is calculated to obtain a first remote control accuracy. The remote control accuracy is an index for quantifying the response accuracy of the remote control function, and has a value range of 0-100%, reflecting the degree of fit between the instruction response and the preset result. The key evaluation dimensions of the preset control result are first disassembled, and a weight is set for each dimension; then the degree of fit between the first remote control result and the preset result in each dimension is compared, and a weighted average score is calculated according to the weight, and the average score is the remote control result similarity, which is directly used as the first remote control accuracy. The specific formula is: first remote control accuracy = Σ (a certain dimension fit score x the weight of the dimension) x 100%. Wherein, the fit score is quantified: complete fit = 1 point, partial fit = 0.5 points, complete mismatch = 0 points. For example, the preset control instruction is: complete TA1 overcurrent protection setting value commissioning within 10 seconds, and the measurement and control system displays "commissioning success"; the first remote control result is: 12 seconds of response, and the state display is "commissioning success". The evaluation dimensions and weights are: execution state weight 0.6; response time weight 0.4. The fit degree is determined: the execution state is completely fit, and 1 point is obtained; the response time of 12 seconds exceeds the standard, and 0.5 points are obtained. The calculation result is: first remote control accuracy = 1 x 0.6 + 0.5 x 0.4 x 100% = (0.6 + 0.2) x 100% = 80%.

[0024] In the embodiment of the application, by confirming the correctness of the hardware foundation first, and then selecting the reference point for testing, the problem of invalid test caused by hardware misconnection is eliminated from the source; at the same time, by using clear standards and quantitative accuracy calculation, accurate initial reference data is provided for the iterative test of the associated points, the accuracy deviation caused by the test without reference is avoided, and the basic accuracy of the substation point information checking is preliminarily ensured.

[0025] S200: According to the moving window, the point electrically connected with the first point is selected for iterative remote measurement test and remote control test, and a remote measurement accuracy matrix and a remote control accuracy matrix are constructed.

[0026] In the embodiment of the application, according to the moving window, the point electrically connected with the first point is selected for iterative remote measurement test and remote control test, and a remote measurement accuracy matrix and a remote control accuracy matrix are constructed. The informationized points in the substation do not exist in isolation, but form a close correlation through the electrical circuit, and the abnormal remote measurement / remote control accuracy of a certain point may be caused by the fault of the point directly connected therewith, and the quality of the overall point network cannot be reflected by testing only a single first point. Therefore, all the associated points are iteratively tested by taking the direct electrical connection as the core and using the moving window, and the scattered accuracy data is matrix integrated according to the position coordinates, so that the one-sidedness of isolated test is avoided, and structured data support is provided for the subsequent accuracy checking based on the topological correlation.

[0027] The step S200 in the method provided by the embodiment of the application comprises: According to the mobile window of direct electrical connection, the point electrically connected with the first point is selected as the plurality of second points; The plurality of second points are tested by remote measurement and remote control, and a plurality of second remote measurement accuracies and a plurality of second remote control accuracies are obtained; The remote measurement accuracy and the remote control accuracy of all the points in the transformer substation are continuously tested, and a remote measurement accuracy matrix and a remote control accuracy matrix are constructed based on the position coordinates of all the points.

[0028] First, according to the mobile window of direct electrical connection, the point electrically connected with the first point is selected as the plurality of second points. The mobile window of direct electrical connection refers to a dynamic screening range containing only the directly connected point of the conductor with the current test point as the core, to ensure that the test object is physically and electrically associated with the initial point. The second point refers to the associated point directly electrically connected with the first point, which is the initial extension object of iterative testing. With the first point as the initial center, the mobile window of direct electrical connection is defined, that is, only the point directly connected with the current point without intermediate electrical elements is screened, and all the information points in this range are selected as the second point; after completing the window test, each second point is taken as a new center, and the window screening logic is repeated until all the points in the transformer substation are covered.

[0029] For example, taking the first point P101 as the initial center, the directly electrically connected point is screened against the secondary wiring diagram: the remote control point P102 of the circuit breaker QF1 directly receives the current signal of the current transformer TA1; the remote measurement point P103 of the disconnecting switch QS1 shares the same power circuit with TA1; the remote measurement point P104 corresponding to the analog input terminal AI2 of the measurement and control device directly collects the secondary side signal of TA1; the above P102, P103 and P104 are the first batch of second points, and after completing the test, P102 is taken as a new center to screen P105 directly connected therewith, and the iteration screening is continued in this way. Through the direct electrical connection window, the associated point is accurately locked, and the test efficiency is avoided from being disturbed by irrelevant points; the iteration logic ensures that all the points are covered without omission, while the topological association relationship between the points is preserved.

[0030] Secondly, the plurality of second points are tested by remote measurement and remote control, and a plurality of second remote measurement accuracies and a plurality of second remote control accuracies are obtained. The second remote measurement accuracy / remote control accuracy: the quantitative index of the remote measurement / remote control accuracy of the second point, the calculation logic is completely consistent with that of the first point, to ensure the comparability of the data. The test standard and accuracy calculation logic of S100 are followed, for each second point, the industry test standard is queried according to the model of the corresponding electrical element, the adaptive preset variable and preset control instruction are set, the test is performed, the second remote measurement variable / remote control result is recorded, and finally the remote measurement accuracy and the remote control accuracy of each second point are calculated through the same formula.

[0031] For example, take the second point P102 (QF1 breaker, function: remote opening / remote opening current) as an example. According to the electrical component test standard, a preset variable opening current of 10A is injected into P102, and the actual second remote variable 9.8A is collected. The second remote accuracy is calculated as (1-|9.8-10| / 10) x 100%=98%; remote control test: send a preset control instruction: complete QF1 opening within 15 seconds, and the measurement and control system displays "opening success". The actual result is 13 seconds opening, and the display is "opening". The second remote control accuracy is calculated as (0.5 x 0.6 + 1 x 0.4) x 100%=70%; similarly, P103 and P104 are tested, and the results are recorded: P103 (second remote accuracy 99%, second remote control accuracy 90%), P104 (second remote accuracy 99.3%, second remote control accuracy, marked "only remote"). The unified test and calculation standard is used to ensure the consistency of the accuracy data of all points, and to provide reliable data basis for subsequent matrix integration and cross-point analysis.

[0032] Finally, the remote measurement accuracy and remote control accuracy of all points in the substation are obtained by continuous testing, and the remote measurement accuracy matrix and the remote control accuracy matrix are constructed based on the position coordinates of all points. The remote measurement accuracy matrix refers to a two-dimensional structured data set indexed by position coordinates, which stores the remote measurement accuracy of all points according to the index relationship of cabinet number+layer number-point code, and directly reflects the remote measurement quality distribution of different position points. The remote control accuracy matrix refers to a structure identical to the remote measurement accuracy matrix, which only stores the remote control accuracy data of all points, and the points without remote control function are marked "-" to distinguish. A moving window is expanded around each measured point, covering all 120 information points in the substation, and the remote measurement and remote control accuracy of all points are calculated. The cabinet number-layer number-terminal number three-dimensional position coordinates of each point are extracted, and a two-dimensional matrix structure is constructed, taking the cabinet number+layer number as the row index of the matrix, such as C10-L1, C10-L2, C10-L3, corresponding to the 1-3 layers of C10 cabinet, and taking the point code as the column index of the matrix, such as P101-P115, corresponding to the 15 points of C10 cabinet. The remote measurement accuracy value of each point is filled in the row and column intersection position of the remote measurement accuracy matrix, and the remote control accuracy value of each point is filled in the corresponding position of the remote control accuracy matrix; only the points with remote measurement function and without remote control function are marked "-" in the corresponding position of the remote control accuracy matrix, and the function attribute is clear.

[0033] For example, after completing all 120 point position tests, focus on the C10 cabinet, which contains 15 point positions, cabinet number C10, layer number L1-L3, and point position code P101-P115. Set the row index of the telemetry accuracy matrix to C10-L1, C10-L2, and C10-L3, and the column index to P101 to P115. Fill in the data according to the index: fill in the telemetry accuracy of P101 at the intersection of the C10-L2 row and the P101 column, fill in the telemetry accuracy of P102 at the intersection of the C10-L2 row and the P102 column, fill in the telemetry accuracy of P103 at the intersection of the C10-L1 row and the P103 column, and fill in the telemetry accuracy of P104 at the intersection of the C10-L3 row and the P104 column. Fill in the remote control accuracy matrix with the same index logic: fill in 80% at the intersection of the C10-L2 row and the P101 column, fill in 70% at the intersection of the C10-L2 row and the P102 column, fill in 90% at the intersection of the C10-L1 row and the P103 column, and mark "-" at the intersection of the C10-L3 row and the P104 column. Complete the matrix data filling of cabinets C11 to C19 according to the above logic, and finally form the telemetry accuracy matrix and the remote control accuracy matrix covering all point positions in the entire substation.

[0034] In the embodiments of the present application, the iterative test logic of the directly electrically connected mobile window ensures that all point positions are covered while the electrical topology association relationship is preserved, avoiding the omission of hidden dangers caused by isolated testing. The unified standard test process guarantees the consistency of the accuracy data. The double-precision matrix construction based on position coordinates converts scattered data into structured assets, which not only realizes the visual management of point position quality, but also provides core data support for subsequent topology matching and accuracy verification of the same family point positions, effectively improving the efficiency and accuracy of subsequent analysis.

[0035] S300: Call the historical point position checking data of the same family point positions, perform accuracy verification on the telemetry accuracy matrix and the remote control accuracy matrix, and obtain a test reliability matrix.

[0036] In the embodiments of the present application, the historical point position checking data of the same family point positions is called to perform accuracy verification on the telemetry accuracy matrix and the remote control accuracy matrix, and a test reliability matrix is obtained. The point positions directly electrically connected in the substation have strong correlation, and the accuracy change of a point position will be transmitted to the associated point positions through the electrical circuit, forming a characteristic accuracy change rule. The existing test only obtains the accuracy of a single point position, and does not verify whether the change amplitude of the associated point position conforms to the conventional rule, which may cause the accuracy data to be distorted due to accidental interference. Therefore, the accuracy change amplitude of the associated point position is calculated, and the historical data of the same family point positions with the same connection topology are verified to judge the rationality of the current accuracy data, and finally a test reliability matrix is formed, providing reliable data quality evaluation basis for subsequent AI fusion identification.

[0037] The step S300 in the method provided by the embodiment of the application comprises: According to the telemetry accuracy matrix and the remote control accuracy matrix, the change amplitudes of the telemetry accuracy and the remote control accuracy of the first point and the plurality of second points are calculated to obtain a plurality of first associated telemetry change amplitudes and a plurality of first associated remote control change amplitudes. A connection topology of the first point and the plurality of second points is constructed, and in the historical point checking data of the other substations, the same family points having the same connection topology are extracted, and the plurality of first associated telemetry change amplitudes and the plurality of first associated remote control change amplitudes are verified and analyzed to obtain a first test credibility. The telemetry accuracy and the remote control accuracy of the other points are continuously checked for accuracy to obtain the test credibility of all the points, and a test credibility matrix is constructed.

[0038] Firstly, according to the telemetry accuracy matrix and the remote control accuracy matrix, the change amplitudes of the telemetry accuracy and the remote control accuracy of the first point and the plurality of second points are calculated to obtain a plurality of first associated telemetry change amplitudes and a plurality of first associated remote control change amplitudes. The first associated telemetry change amplitude is the absolute difference value of the telemetry accuracy of the second point and the first point, reflecting the fluctuation degree of the telemetry accuracy between the associated points. The first associated remote control change amplitude is the absolute difference value of the remote control accuracy of the second point and the first point, reflecting the fluctuation degree of the remote control accuracy between the associated points. Taking the accuracy data of the first point as a reference, the telemetry accuracy difference value and the remote control accuracy difference value of each second point and the first point are calculated, and the absolute value of the difference value is the associated accuracy change amplitude, corresponding to the first associated telemetry change amplitude and the first associated remote control change amplitude respectively; the point without remote control function only calculates the telemetry change amplitude.

[0039] For example, it is known that the telemetry accuracy of the first point P101 is 99.5%, and the remote control accuracy is 80%, and the accuracy data of the second point is as follows: P102 (telemetry accuracy 98%, remote control accuracy 70%): first associated telemetry change amplitude = |98%-99.5%|=1.5%; first associated remote control change amplitude = |70%-80%|=10%; P103 (telemetry accuracy 99%, remote control accuracy 90%): first associated telemetry change amplitude = |99%-99.5%|=0.5%; first associated remote control change amplitude = |90%-80%|=10%; P104 (telemetry accuracy 99.3%, without remote control): first associated telemetry change amplitude = |99.3%-99.5%|=0.2%; first associated remote control change amplitude is marked as “-”; finally, the first associated telemetry change amplitude set {1.5%, 0.5%, 0.2%} is obtained, and the first associated remote control change amplitude set {10%, 10%, -} is obtained. By quantifying the accuracy change amplitudes of the associated points, the implicit electrical association between the points is converted into intuitive data, providing core analysis indexes for subsequent comparison with the same family points.

[0040] Secondly, a connection topology of the first point and the plurality of second points is constructed, and in historical point checking data of other substations, same family points having the same connection topology are extracted, the plurality of first associated telemetry change amplitudes and the plurality of first associated remote control change amplitudes are verified and analyzed, and a first test credibility is obtained.

[0041] The connection topology of the first point and the plurality of second points is constructed, and in historical point checking data of other substations, same family points having the same connection topology are extracted, the plurality of first associated telemetry change amplitudes and the plurality of first associated remote control change amplitudes are verified and analyzed, and a first test credibility is obtained, and the method comprises the following steps: The connection topology of the first point and the plurality of second points is constructed, and in historical point checking data of other substations, same family points having the same connection topology are extracted; The associated telemetry change amplitude mean and the associated remote control change amplitude mean of the same family points in the telemetry and remote control test with the connection point are obtained as the plurality of same family associated telemetry change amplitudes and the plurality of same family associated remote control change amplitudes; The similarity of the plurality of first associated telemetry change amplitudes and the plurality of same family associated telemetry change amplitudes is calculated respectively, and a plurality of telemetry test credibilities are obtained; The similarity of the plurality of first associated remote control change amplitudes and the plurality of same family associated remote control change amplitudes is calculated respectively, and a plurality of remote control test credibilities are obtained; Based on the plurality of telemetry test credibilities and the plurality of remote control test credibilities, a first test credibility of the first point is calculated and obtained.

[0042] Firstly, a connection topology of the first point and the plurality of second points is constructed, and in historical point checking data of other substations, same family points having the same connection topology are extracted. The connection topology refers to a structure diagram directly showing the electrical connection path of the first point and the second point, and clearly showing the association relationship between the points. The same family points refer to a point set in the historical database, which is completely consistent with the current point group in voltage level, element combination and connection topology, and the associated change rule has reference value. Based on the direct electrical connection relationship between the first point and the plurality of second points, the connection path of the first point and the second point is clearly shown in a graphical manner, and the connection topology is formed; then the historical database is searched, and other substation point sets of the same voltage level, the same element combination and the same connection topology structure are screened as the same family points.

[0043] Exemplary, construct the connection topology: take P101 (TA1) as the core, draw the topology map: P101 (TA1) is directly connected to P102 (QF1), P101 (TA1) is directly connected to P103 (QS1), P101 (TA1) is directly connected to P104 (AI2), and 3 groups of direct connection relationships are clear. Extract the same family point: from the historical database, such as the data of 50 110kV substations, filter out 12 groups of point groups that meet the 10kV voltage level, the current transformer TA+ circuit breaker QF+ disconnecting switch QS+ measurement and control device element combination, and the same connection topology as described above, as the same family point corresponding to P101. The connection topology locks the association core of the current point group, and the same family point screening ensures the comparability of the subsequent comparison data, providing a reliable reference benchmark for rule verification.

[0044] Secondly, the average of the associated remote measurement change range and the average of the associated remote control change range in the remote measurement and remote control test of the same family point are obtained, as the multiple same family associated remote measurement change ranges and the multiple same family associated remote control change ranges. The same family associated remote measurement change range refers to the average of the historical associated remote measurement change range corresponding to the connection relationship in the same family point group. The same family associated remote control change range refers to the average of the historical associated remote control change range corresponding to the connection relationship in the same family point group. For each corresponding connection relationship in the same family point group, the historical remote measurement / remote control associated change range data is extracted, the average of each corresponding relationship is calculated, and the same family associated remote measurement change range and the same family associated remote control change range are obtained.

[0045] Exemplary, for the 3 groups of corresponding connection relationships of the same family point, the historical associated change range of each group is extracted: TA-QF: historical remote measurement change range data {1.1%, 1.3%, 1.2%, …}, the average is 1.2%; historical remote control change range data {7.8%, 8.2%, 8.0%, …}, the average is 8%; TA-QS: the average of the historical remote measurement change range is 0.6%, and the average of the historical remote control change range is 12%; TA-AI: the average of the historical remote measurement change range is 0.3%, and the remote control function is marked as “-”; finally, the same family associated remote measurement change range set {1.2%, 0.6%, 0.3%} and the same family associated remote control change range set {8%, 12%, -} are obtained. Through the average calculation, the common change rule of the same family point is extracted, a quantitative benchmark for the rationality judgment of the current change range is provided, and accidental interference of single group of historical data is avoided.

[0046] Further, the similarity of the plurality of first associated telemetry change amplitudes and the plurality of same-family associated telemetry change amplitudes is calculated respectively to obtain a plurality of telemetry test reliabilities. The telemetry test reliability is the degree of fit of the telemetry change amplitude of a single set of associated points with the same-family mean value, and the value is 0-100%. The higher the similarity, the more the current telemetry change conforms to the electrical association rule, and the lower the possibility of accidental error. Similarity = 1- | current change amplitude - same-family mean value | / same-family mean value x 100%. The similarity of each first associated telemetry change amplitude and the corresponding same-family associated telemetry change amplitude is calculated respectively, and each similarity is the telemetry test reliability of the corresponding connection relationship.

[0047] For example, the known first associated telemetry change amplitudes {1.5% (P101-P102), 0.5% (P101-P103), 0.2% (P101-P104)} and the same-family mean values {1.2%, 0.6%, 0.3%}: P101-P102: Similarity = 1- | 1.5%-1.2% | / 1.2% x 100% ≈ 97.5%, i.e. telemetry test reliability 97.5%; P101-P103: Similarity = 1- | 0.5%-0.6% | / 0.6% x 100% ≈ 98.3%, i.e. telemetry test reliability 98.3%; P101-P104: Similarity = 1- | 0.2%-0.3% | / 0.3% x 100% ≈ 96.7%, i.e. telemetry test reliability 96.7%; and finally a plurality of telemetry test reliability sets {97.5%, 98.3%, 96.7%} are obtained. The similarity is quantified to verify the rule of the telemetry change amplitude, intuitively distinguish the case of conforming to the electrical association and the possible accidental error, and provide a subdivision basis for the overall reliability.

[0048] At the same time, the similarity of the plurality of first associated remote control change amplitudes and the plurality of same-family associated remote control change amplitudes is calculated respectively to obtain a plurality of remote control test reliabilities. The remote control test reliability refers to the degree of fit of the remote control change amplitude of a single set of associated points with the same-family mean value, and the value is 0-100%. It reflects the rationality of the current remote control change. The calculation logic of the telemetry test reliability is followed. For each first associated remote control change amplitude and the corresponding same-family associated remote control change amplitude, the same similarity formula is used to calculate the degree of fit to obtain the corresponding remote control test reliability of each group. The points without remote control function are marked as “-” and do not participate in the calculation.

[0049] For example, the known first associated remote control change amplitude {10% (P101-P102), 10% (P101-P103), -}, the average change amplitude of the associated remote control of the same family {8%, 12%, -}: P101-P102: similarity = 1- |10%-8%| / 8% x 100% = 97.5%, i.e. the remote control test reliability is 97.5%; P101-P103: similarity = 1- |10%-12%| / 12% x 100% ≈ 98.3%, i.e. the remote control test reliability is 98.3%; finally, a plurality of remote control test reliability sets {97.5%, 98.3%} are obtained. Complementary to the telemetry test reliability, the telemetry and remote control functions of the associated points are comprehensively covered, and the rationality of the accuracy data of the two types is verified.

[0050] Finally, based on the plurality of telemetry test reliabilities and the plurality of remote control test reliabilities, the first test reliability of the first point is calculated. The first test reliability refers to the verification result of the law of a plurality of associated points of telemetry and remote control, and is the final index of the reliability of the accuracy data of the first point, taking values 0-100%. The weights of the telemetry test reliability and the remote control test reliability are both set to 0.5, the average value of the plurality of telemetry test reliabilities and the average value of the plurality of remote control test reliabilities are calculated first, and then the weighted sum is calculated according to the weights, and the result is the first test reliability of the first point.

[0051] For example, the average telemetry reliability is calculated: (97.5%+98.3%+96.7%) / 3 ≈ 97.5%; the average remote control reliability is calculated: (97.5%+98.3%) / 2 ≈ 97.9%; the first test reliability is calculated by weighting: (97.5% x 0.5+97.9% x 0.5) x 100% ≈ 97.7%. Through the average and weighting of a plurality of sub reliabilities, the correlation law consistency of telemetry and remote control is comprehensively considered, the influence of single data deviation is avoided, and the first test reliability is more objective and persuasive.

[0052] Finally, continue to check the telemetry accuracy and remote control accuracy of other points, and obtain the test reliability of all points to construct a test reliability matrix. The test reliability matrix refers to a structured data set that stores the reliability indicators of all points after the same family topology verification, which is completely consistent with the structure of the telemetry / remote control accuracy matrix. The verification logic of the first point is followed, and each point in the substation that is not the core is taken as a new core point. The complete process of screening associated points with direct electrical connection, calculating the telemetry / remote control associated change amplitude of the core and the associated points, constructing the connection topology of the core point, extracting the average historical change amplitude of the same family points, calculating the test reliability of multiple groups of telemetry / remote control, and weighting to obtain the test reliability of the core point is sequentially executed. After the test reliability of all points is calculated, according to the unified index structure of the telemetry / remote control accuracy matrix in S200, the test reliability of each point is filled in the corresponding position, and the reliability of the point without remote control function is calculated separately according to the telemetry test reliability. Finally, the test reliability matrix covering the entire substation is formed.

[0053] For example, the new core point P102 (QF1) is iteratively verified: associated points: P101 (TA1), P105 (QF1 auxiliary contact point); accuracy data: P102 (telemetry accuracy 98%, remote control accuracy 70%), P101 (telemetry accuracy 99.5%, remote control accuracy 80%), P105 (telemetry accuracy 99.2%, remote control accuracy 85%); associated change amplitude: P102-P101 (associated telemetry change amplitude 1.5%, associated remote control change amplitude 10%), P102-P105 (associated telemetry change amplitude 1.2%, associated remote control change amplitude 15%); connection topology: P102-P101, P102-P105; average associated remote control change amplitude of the same family: P102-P101 (telemetry 1.3%, remote control 9%), P102-P105 (telemetry 1.0%, remote control 14%); test reliability: telemetry average reliability 98.25%, remote control average reliability 99.1%, final reliability ≈98.7%; similarly, other points are iteratively verified: P103 (QS1): associated P101, P106, test reliability 97.9%; P104 (AI2): only telemetry, associated P101, test reliability 98.3%; P105: associated P102, test reliability 97.5%; construct matrix (C10 cabinet as an example): C10-L2 row: P101 (97.7%), P102 (98.7%), P105 (97.5%); C10-L1 row: P103 (97.9%); C10-L3 row: P104 (98.3%); complete C11-C19 cabinet point filling, and form a test reliability matrix covering 120 points.

[0054] In the embodiment of the present application, the isolated point accuracy data is converted into reliable data verified by correlation rules by calculating the correlation change amplitude, topological matching homologous points, and verifying the similarity, effectively filtering the distorted data caused by accidental interference; the test reliability matrix is unified with the previous double-precision matrix structure, realizing one-to-one correspondence of data quality and accuracy data, providing a key basis for weight allocation in subsequent AI fusion identification, and further improving the accuracy of substation point information checking results.

[0055] S400: According to the test reliability matrix, the telemetry accuracy matrix and the remote control accuracy matrix, AI fusion identification is performed to obtain a point checking quality parameter matrix as a substation point information checking result.

[0056] In the embodiment of the present application, according to the test reliability matrix, the telemetry accuracy matrix and the remote control accuracy matrix, AI fusion identification is performed to obtain a point checking quality parameter matrix as a substation point information checking result. The telemetry accuracy and the remote control accuracy are two core dimensions reflecting the quality of the substation point, the former reflects the parameter collection accuracy, and the latter reflects the instruction response reliability. Single-dimensional data cannot fully represent the real operation quality of the point. AI fusion identification can deeply mine the internal correlation characteristics of the two types of accuracy data, output more comprehensive basic quality evaluation; and the test reliability matrix has been verified by homologous topology and can accurately reflect the reliability of the accuracy data. The test reliability matrix is used to modify the basic evaluation result, filter the distorted data caused by accidental errors, make the final point checking quality parameter more in line with the actual situation, and further improve the accuracy and reliability of the checking result.

[0057] The step S400 in the method provided by the embodiment of the present application includes: According to the telemetry accuracy matrix and the remote control accuracy matrix, AI fusion identification is performed to obtain a basic point checking quality parameter matrix. According to the test reliability matrix, the basic point checking quality parameter matrix is modified to obtain a point checking quality parameter matrix as a substation point information checking result.

[0058] First, according to the telemetry accuracy matrix and the remote control accuracy matrix, AI fusion identification is performed to obtain a basic point checking quality parameter matrix.

[0059] According to the telemetry accuracy matrix and the remote control accuracy matrix, AI fusion identification is performed to obtain a basic point checking quality parameter matrix, including: According to the substation point information checking record in the historical time, a sample telemetry accuracy set and a sample remote control accuracy set are collected, and a real point electrical quality parameter verified by checking is collected to label a sample point checking quality parameter set. Based on AI machine learning, a point checking analysis network is constructed. The sample telemetry accuracy set, the sample remote control accuracy set and the sample point checking quality parameter set are used to supervise training and testing of the point checking analysis network until the testing accuracy rate is qualified. The telemetry accuracy and the remote control accuracy of the points in the telemetry accuracy matrix and the remote control accuracy matrix are combined and input into the point checking analysis network to output basic point checking quality parameters and obtain a basic point checking quality parameter matrix.

[0060] First, according to the substation point checking records in the historical time, the sample telemetry accuracy set and the sample remote control accuracy set are collected, and the real point electrical quality parameters after checking and verification are collected to obtain the sample point checking quality parameter set. The sample telemetry accuracy set refers to the telemetry accuracy data set of the historical points, which is used for the AI model to learn the correlation between the telemetry accuracy and the real quality. The sample remote control accuracy set refers to the remote control accuracy data set of the historical points, which is used for the AI model to learn the correlation between the remote control accuracy and the real quality. The sample point checking quality parameter set refers to the sample set with real quality parameters labeled, which is used for the supervised training of the AI model. From the point checking records of multiple substations of the same voltage level in the historical time, the sample telemetry accuracy and the sample remote control accuracy of all points are collected to form the sample telemetry accuracy set and the sample remote control accuracy set; at the same time, the real electrical quality parameters of these points after artificial review or device operation verification are extracted as sample point checking quality parameters, which are labeled to form the sample point checking quality parameter set.

[0061] For example, from the historical records of 50 110kV substations, 12000 point sample data are collected: the sample telemetry accuracy set includes C10 cabinet P101 historical same period telemetry accuracy 99.2%, P102 historical telemetry accuracy 97.8%, etc.; the sample remote control accuracy set includes C10 cabinet P101 historical same period remote control accuracy 82%, P102 historical remote control accuracy 71%, etc.; the sample point checking quality parameter set is labeled with P101 historical real quality parameter 95 points, P102 historical real quality parameter 90 points, etc., and the higher the score, the better the point quality. The AI model is provided with sufficient training samples to ensure that the model can learn the internal correlation rules between telemetry, remote control accuracy and real point quality.

[0062] Secondly, based on AI machine learning, a point checking analysis network is constructed. The point checking analysis network refers to an AI model specially used for integrating telemetry and remote control precision features to output point quality parameters. Based on AI machine learning technology, a neural network structure suitable for numerical feature fusion, such as a multilayer perceptron, is selected to construct the point checking analysis network. The input layer is a two-dimensional feature vector of single-point telemetry precision + single-point remote control precision, the hidden layer extracts and fuses features through an activation function, and the output layer is a single numerical point checking quality parameter, 0-100 points, with the higher the score, the better the quality. An AI model architecture adapted to double-precision fusion is built, laying a foundation for subsequent accurate output of quality parameters. For example, a 3-layer neural network is constructed: the input layer has 2 neurons, which receive telemetry precision and remote control precision respectively; the hidden layer has 16 neurons, which use a ReLU activation function; the output layer has 1 neuron, which outputs a quality parameter of 0-100 points; and the core function of the network is to learn the internal correlation between telemetry, remote control precision and point true quality.

[0063] Further, the sample telemetry precision set, the sample remote control precision set and the sample point checking quality parameter set are used to supervise the training and testing of the point checking analysis network until the test accuracy rate is qualified. Supervised training refers to training a model with sample data with true labels to let the model learn the explicit association between input and output. The test accuracy rate reflects the degree of fit between the predicted quality parameter and the true quality parameter of the model, and the higher the accuracy rate, the better the model performance. The sample telemetry precision set and the sample remote control precision set are combined in the format of a single-point two-dimensional feature vector as the model input, and the sample point checking quality parameter set is used as the model output label for supervised training. The network parameters are adjusted through back propagation to reduce the error between the predicted value and the true label. The model accuracy rate is verified every 10 training rounds with a test set until the test accuracy rate is ≥95%, the training is stopped and the model is saved.

[0064] For example, 6000 sample data are used as the training set and 1500 sample data are used as the test set. The initial model test accuracy rate is 72%, and after 50 rounds of training, the test accuracy rate stabilizes at 96.3%, which is ≥95% of the qualified standard. The training of the point checking analysis network is stopped and the trained network is saved. By training and optimizing the model parameters, it is ensured that the model can accurately mine the association between double-precision data and point quality, providing a reliable basis for outputting quality parameters.

[0065] Subsequently, the telemetry accuracy and the remote control accuracy of a unified point position in the telemetry accuracy matrix and the remote control accuracy matrix are combined, input into the point position checking analysis network, and a basic point position checking quality parameter is output to obtain a basic point position checking quality parameter matrix. The basic point position checking quality parameter refers to a preliminary point position quality evaluation value output by the AI model after integrating the telemetry accuracy and the remote control accuracy, and the value is 0-100 points. The basic point position checking quality parameter matrix refers to a structured data set storing all basic quality parameters of the point positions according to a unified index, and the structure is consistent with the previous matrix. The telemetry accuracy and the remote control accuracy of each point position are extracted from the telemetry accuracy matrix and the remote control accuracy matrix, combined into a two-dimensional feature vector of a unified point position, the feature vectors of all point positions are sequentially input into the trained point position checking analysis network, and a basic point position checking quality parameter of each point position is output. The basic quality parameters are filled into the corresponding positions according to a unified index structure of cabinet number + layer number-point position code, and the basic point position checking quality parameter matrix is constructed. Through the AI model integrating the double-precision data, the scattered accuracy indicators are converted into unified quality parameters, and comprehensive preliminary evaluation of the point position quality is realized to provide basic data for subsequent correction.

[0066] For example, the two-dimensional feature vectors of the C10 cabinet point positions are extracted: P101: telemetry accuracy 99.5%, remote control accuracy 80%; P102: telemetry accuracy 98%, remote control accuracy 70%; P103: telemetry accuracy 99%, remote control accuracy 90%; P104: telemetry accuracy 99.3%, remote control function missing; P105: telemetry accuracy 99.2%, remote control accuracy 85%; the same are input into the trained point position checking analysis network, and the basic quality parameters are output: P101: 94 points, P102: 88 points, P103: 95 points, P104: 92 points, P105: 93 points; the basic matrix is constructed according to the index, and the C10 cabinet is taken as an example: C10-L2 row: P101 (94 points), P102 (88 points), P105 (93 points); C10-L1 row: P103 (95 points); C10-L3 row: P104 (92 points).

[0067] Secondly, according to the test reliability matrix, the basic point position checking quality parameter matrix is corrected to obtain a point position checking quality parameter matrix as a substation point position information checking result.

[0068] According to the test reliability matrix, the basic point position checking quality parameter matrix is corrected to obtain a point position checking quality parameter matrix as a substation point position information checking result, including: According to the test reliability matrix, a point position checking quality correction coefficient matrix is configured; The point position checking quality correction coefficient matrix is used to correct and calculate the basic point position checking quality parameter matrix to obtain a point position checking quality parameter matrix as a substation point position information checking result.

[0069] First, according to the test credibility matrix, the point position check quality correction coefficient matrix is configured. The point position check quality correction coefficient refers to the adjustment coefficient set according to the test credibility, which is used to correct the basic quality parameter and reflects the influence of the reliability of the accuracy data on the quality evaluation. The point position check quality correction coefficient matrix is consistent with the previous matrix structure, and stores a structured data set of all point position correction coefficients. According to the test credibility of each point position in the test credibility matrix, the corresponding point position check quality correction coefficient is set. The higher the test credibility, the closer the correction coefficient to 1, indicating that the basic quality parameter is more reliable and does not need to be adjusted significantly. The lower the test credibility, the farther the correction coefficient deviates from 1, indicating that the basic quality parameter may be distorted and needs to be adjusted. For example, the rules are set as follows: credibility ≥ 98%, correction coefficient 1.0; 95%≤ credibility < 98%, correction coefficient 0.99; 90%≤ credibility < 95%, correction coefficient 0.98; credibility < 90%, correction coefficient 0.95. According to the index of cabinet number + layer number-point position code, the correction coefficient is filled into the corresponding position to construct the point position check quality correction coefficient matrix. Or directly use the test credibility as the point position check quality correction coefficient, for example, if the credibility is 98%, the correction coefficient is 0.98. The test credibility is converted into a quantifiable correction coefficient, which provides a clear basis for accurate adjustment of the basic quality parameter and ensures that the correction logic is traceable.

[0070] For example, the test credibility of the C10 cabinet point position is extracted: P101: 97.7%, P102: 98.7%, P103: 97.9%, P104: 98.3%, P105: 97.5%; according to the rules, the correction coefficient is configured: P101 (97.7%): 0.99, P102 (98.7%): 1.0, P103 (97.9%): 0.99, P104 (98.3%): 1.0, P105 (97.5%): 0.99; the correction coefficient matrix is constructed (C10 cabinet is taken as an example): C10-L2 row: P101 (0.99), P102 (1.0), P105 (0.99); C10-L1 row: P103 (0.99); C10-L3 row: P104 (1.0).

[0071] Secondly, the point position checking quality correction coefficient matrix is used to correct and calculate the basic point position checking quality parameter matrix to obtain a point position checking quality parameter matrix as a substation point position information checking result. The final point position checking quality parameter refers to a point position quality evaluation value corrected by the test reliability, and is more suitable for the real operation quality of the point position. The point position checking quality parameter matrix is a structured data set storing final quality parameters of all point positions, and is a core achievement of this checking. The final point position checking quality parameter = basic point position checking quality parameter * point position checking quality correction coefficient, and each point position data in the basic point position checking quality parameter matrix is corrected and calculated. The final quality parameter after correction is filled in the corresponding position according to the unified index structure of cabinet number + layer number-point position code to form the point position checking quality parameter matrix as the final result of the substation point position information checking.

[0072] For example, the final quality parameters of C10 cabinet point positions are corrected and calculated as follows: P101: 94 points * 0.99 = 93 points; P102: 88 points * 1.0 = 88 points; P103: 95 points * 0.99 = 94 points; P104: 92 points * 1.0 = 92 points; P105: 93 points * 0.99 = 92 points. The final matrix is constructed, and C10 cabinet is taken as an example: C10-L2 row: P101 (93 points), P102 (88 points), P105 (92 points); C10-L1 row: P103 (94 points); C10-L3 row: P104 (92 points). The correction calculation and matrix filling of all point positions of C11-C19 cabinets are sequentially completed to finally form the point position checking quality parameter matrix covering 120 point positions of the whole substation. The final quality parameter is closer to the real state of the point position by filtering the evaluation deviation caused by the precision data distortion through the correction coefficient, and the reliability of the checking result is ensured.

[0073] In the embodiment of the application, the two types of precision data of telemetry and remote control are integrated by AI fusion identification, the one-dimension evaluation is broken through, the preliminary evaluation of the point position quality is realized, the distortion data caused by accidental errors is effectively filtered according to the test reliability matrix, and the final point position checking quality parameter matrix is comprehensive and accurate. The point position checking quality parameter matrix is unified with the previous matrix structure, the real quality level of each point position of the whole substation can be directly presented, the clear checking result is provided for the operation and maintenance personnel, and the accuracy and efficiency of the substation point position information checking are finally effectively improved.

[0074] The embodiment of the application achieves the following technical effects through the specific implementation manner. The application provides a substation site information checking method and system based on AI recognition fusion, a reliable origin is established by performing single-point benchmark testing; secondly, a precision matrix of a covering network is constructed by testing an electrically associated point group through mobile window iteration, and the test efficiency is improved; then, cross-checking is performed on the same family historical data to generate a credibility matrix, effectively suppressing accidental errors and enhancing data reliability; finally, AI fusion technology is used to intelligently analyze and weight distribute multiple source matrices, and a comprehensive quantitative quality parameter matrix is output. The application realizes the transformation from isolated single-point testing to network evaluation and from manual judgment to intelligent decision-making, improves the test automation degree and efficiency, and improves the accuracy and precision of the evaluation results through multi-dimensional checking and fusion analysis.

[0075] In an embodiment, as shown in FIG. 1, the application provides a substation site information checking system based on AI recognition fusion, which comprises: Figure 2 a first point precision test module 11, configured to, after the hardware connection checking in the substation site information checking is completed, select a first point for remote measurement testing and remote control testing, and obtain a first remote measurement precision and a first remote control precision; an iterative test module 12, configured to select points electrically connected to the first point according to a mobile window, perform iterative remote measurement testing and remote control testing, and construct a remote measurement precision matrix and a remote control precision matrix; a precision checking module 13, configured to call historical point checking data of the same family points to perform precision checking on the remote measurement precision matrix and the remote control precision matrix, and obtain a test credibility matrix; an AI fusion recognition module 14, configured to perform AI fusion recognition according to the test credibility matrix, the remote measurement precision matrix and the remote control precision matrix, and obtain a point checking quality parameter matrix as a substation site information checking result. In one embodiment, the first point precision test module 11 is further configured to:

[0076] after the hardware connection checking in the substation site information checking is completed and the checking result is correct, read a substation design file, and select a first point for remote measurement and remote control; perform remote measurement testing and remote control testing according to the test standard of the electric element corresponding to the first point, and obtain a first remote measurement precision and a first remote control precision. In one embodiment, the first point precision test module 11 is further configured to:

[0077] perform remote measurement testing and remote control testing according to the test standard of the electric element corresponding to the first point, and obtain a first remote measurement precision and a first remote control precision, which comprises: setting a preset variable for the first point according to the test standard of the electric element corresponding to the first point, performing remote measurement testing to obtain a first remote measurement variable, and performing remote control on the first point through a preset control instruction to obtain a first remote control result; calculate a similarity between the first telemetry variable and the preset variable as a first telemetry accuracy; calculate a similarity between the first remote control result and a preset control result in the preset control instruction, to obtain a first remote control accuracy.

[0078] In one embodiment, the iterative test module 12 is further configured to: select, as a plurality of second point locations, point locations electrically connected to the first point location according to a moving window of direct electrical connection; perform telemetry testing and remote control testing on the plurality of second point locations to obtain a plurality of second telemetry accuracies and a plurality of second remote control accuracies; continue to test to obtain telemetry accuracies and remote control accuracies of all point locations in the substation, and construct a telemetry accuracy matrix and a remote control accuracy matrix based on position coordinates of all point locations.

[0079] In one embodiment, the accuracy verification module 13 is further configured to: calculate a change range of the telemetry accuracy and the remote control accuracy of the first point location and the plurality of second point locations according to the telemetry accuracy matrix and the remote control accuracy matrix, to obtain a plurality of first associated telemetry change ranges and a plurality of first associated remote control change ranges; construct a connection topology of the first point location and the plurality of second point locations, extract homologous point locations having the same connection topology in historical point location check data of other substations, and verify and analyze the plurality of first associated telemetry change ranges and the plurality of first associated remote control change ranges, to obtain a first test credibility; continue to perform accuracy verification on the telemetry accuracy and the remote control accuracy of other point locations, to obtain test credibilities of all point locations, and construct a test credibility matrix.

[0080] In one embodiment, the accuracy verification module 13 is further configured to: construct a connection topology of the first point location and the plurality of second point locations, extract homologous point locations having the same connection topology in historical point location check data of other substations, and verify and analyze the plurality of first associated telemetry change ranges and the plurality of first associated remote control change ranges, to obtain a first test credibility; obtain an average associated telemetry change range and an average associated remote control change range of the homologous point locations in the telemetry and remote control testing with the connection point locations, as a plurality of homologous associated telemetry change ranges and a plurality of homologous associated remote control change ranges; respectively calculate a similarity between the plurality of first associated telemetry change ranges and the plurality of homologous associated telemetry change ranges, to obtain a plurality of telemetry test credibilities; Calculate the similarity of the plurality of first associated remote control change amplitudes and the plurality of same-family associated remote control change amplitudes respectively, and obtain a plurality of remote control test reliabilities; Based on the plurality of telemetry test reliabilities and the plurality of remote control test reliabilities, a first test reliability of the first point position is calculated and obtained.

[0081] In one embodiment, the AI fusion identification module 14 is further used to: According to the telemetry accuracy matrix and the remote control accuracy matrix, AI fusion identification is performed to obtain a basic point position checking quality parameter matrix; According to the test reliability matrix, the basic point position checking quality parameter matrix is corrected to obtain a point position checking quality parameter matrix as a substation point position information checking result.

[0082] According to the telemetry accuracy matrix and the remote control accuracy matrix, AI fusion identification is performed to obtain a basic point position checking quality parameter matrix, including: According to the telemetry accuracy matrix and the remote control accuracy matrix, AI fusion identification is performed to obtain a basic point position checking quality parameter matrix, including: Based on AI machine learning, a point position checking analysis network is constructed; The sample telemetry accuracy set, the sample remote control accuracy set, and the sample point position checking quality parameter set are used to supervise training and testing of the point position checking analysis network until the test accuracy rate is qualified; The telemetry accuracy and the remote control accuracy of a common point position in the telemetry accuracy matrix and the remote control accuracy matrix are combined and input into the point position checking analysis network, and a basic point position checking quality parameter is output to obtain a basic point position checking quality parameter matrix.

[0083] According to the test reliability matrix, the basic point position checking quality parameter matrix is corrected to obtain a point position checking quality parameter matrix as a substation point position information checking result, including: According to the test reliability matrix, a point position checking quality correction coefficient matrix is configured; The point position checking quality correction coefficient matrix is used to correct and calculate the basic point position checking quality parameter matrix to obtain a point position checking quality parameter matrix as a substation point position information checking result.

[0084] It should be noted that the above-mentioned embodiment sequences of the present application are merely intended for description only, but not used to represent the advantages and disadvantages of the embodiments. And the above-mentioned embodiments of the present application are described in the present specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0085] The above-mentioned is only the preferred embodiment of the present application, and does not limit the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

[0086] The present specification and drawings are merely exemplary of the present application, and any and all modifications, variations, combinations or equivalents that are within the scope of the present application should be included. Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is intended to include these modifications and variations.

Claims

1. A substation site information checking method based on AI recognition fusion, characterized in that, The method comprises: After the hardware connection check in the substation point information check is completed, a first point is selected for remote measurement test and remote control test to obtain a first remote measurement accuracy and a first remote control accuracy; According to a moving window, a point electrically connected to the first point is selected for iterative remote measurement test and remote control test, and a remote measurement accuracy matrix and a remote control accuracy matrix are constructed; The historical point check data of the same family point is called to perform accuracy verification on the remote measurement accuracy matrix and the remote control accuracy matrix to obtain a test credibility matrix; According to the test credibility matrix, the remote measurement accuracy matrix and the remote control accuracy matrix, AI fusion recognition is performed to obtain a point check quality parameter matrix as a substation point information check result. 2.The AI recognition fusion-based substation site information checking method according to claim 1, characterized in that, After the hardware connection check in the substation point information check is completed, a first point is selected for remote measurement test and remote control test to obtain a first remote measurement accuracy and a first remote control accuracy, comprising: After the hardware connection check in the substation point information check is completed and the check result is correct, a substation design file is read, and a first point for remote measurement and remote control is selected; According to the test standard of the electrical element corresponding to the first point, remote measurement test and remote control test are performed to obtain a first remote measurement accuracy and a first remote control accuracy. 3.The AI recognition fusion-based substation site information checking method according to claim 2, characterized in that, According to the test standard of the electrical element corresponding to the first point, remote measurement test and remote control test are performed to obtain a first remote measurement accuracy and a first remote control accuracy, comprising: According to the test standard of the electrical element corresponding to the first point, a preset variable is set for the first point, remote measurement test is performed to obtain a first remote measurement variable, and remote control of the first point is performed according to a preset control instruction to obtain a first remote control result; The similarity of the first remote measurement variable and the preset variable is calculated as the first remote measurement accuracy; The similarity of the first remote control result and a preset control result in the preset control instruction is calculated to obtain the first remote control accuracy. 4.The AI recognition fusion-based substation site information checking method according to claim 1, characterized in that, According to a moving window, a point electrically connected to the first point is selected for iterative remote measurement test and remote control test, and a remote measurement accuracy matrix and a remote control accuracy matrix are constructed, comprising: According to the moving window of direct electrical connection, the point electrically connected to the first point is selected as a plurality of second points; Remote measurement test and remote control test are performed on the plurality of second points to obtain a plurality of second remote measurement accuracies and a plurality of second remote control accuracies; Continue to test to obtain remote measurement accuracies and remote control accuracies of all points in the substation, and construct a remote measurement accuracy matrix and a remote control accuracy matrix based on the position coordinates of all points. 5.The AI recognition fusion-based substation site information checking method according to claim 1, characterized in that, The historical point check data of the same family point is called to perform accuracy verification on the remote measurement accuracy matrix and the remote control accuracy matrix to obtain a test credibility matrix, comprising: According to the remote measurement accuracy matrix and the remote control accuracy matrix, the change amplitudes of the remote measurement accuracies and the remote control accuracies of the first point and the plurality of second points are calculated to obtain a plurality of first associated remote measurement change amplitudes and a plurality of first associated remote control change amplitudes; A connection topology of the first point and the plurality of second points is constructed, same family points having the same connection topology are extracted from the historical point check data of other substations, and the plurality of first associated remote measurement change amplitudes and the plurality of first associated remote control change amplitudes are verified and analyzed to obtain a first test credibility; Continue to check the telemetry accuracy and remote control accuracy of other points, obtain the test reliability of all points, and construct a test reliability matrix. 6.The AI recognition fusion-based substation site information checking method according to claim 5, characterized in that, Construct a connection topology of the first point and the plurality of second points, extract same-family points with the same connection topology from historical point checking data of other substations, verify and analyze the plurality of first associated telemetry change amplitudes and the plurality of first associated remote control change amplitudes, and obtain a first test reliability, including: Construct a connection topology of the first point and the plurality of second points, extract same-family points with the same connection topology from historical point checking data of other substations; Obtain the average associated telemetry change amplitude and the average associated remote control change amplitude of the same-family points in the telemetry and remote control tests with the connection points as the plurality of same-family associated telemetry change amplitudes and the plurality of same-family associated remote control change amplitudes; Calculate the similarity of the plurality of first associated telemetry change amplitudes and the plurality of same-family associated telemetry change amplitudes respectively, and obtain a plurality of telemetry test reliabilities; Calculate the similarity of the plurality of first associated remote control change amplitudes and the plurality of same-family associated remote control change amplitudes respectively, and obtain a plurality of remote control test reliabilities; Based on the plurality of telemetry test reliabilities and the plurality of remote control test reliabilities, calculate and obtain the first test reliability of the first point. 7.The AI recognition fusion-based substation site information checking method according to claim 1, characterized in that, According to the test reliability matrix, the telemetry accuracy matrix and the remote control accuracy matrix, perform AI fusion identification to obtain a point checking quality parameter matrix as a substation point information checking result, including: According to the telemetry accuracy matrix and the remote control accuracy matrix, perform AI fusion identification to obtain a basic point checking quality parameter matrix; According to the test reliability matrix, correct the basic point checking quality parameter matrix to obtain a point checking quality parameter matrix as a substation point information checking result. 8.The AI recognition fusion-based substation site information checking method according to claim 7, characterized in that, According to the telemetry accuracy matrix and the remote control accuracy matrix, perform AI fusion identification to obtain a basic point checking quality parameter matrix, including: According to the substation point information checking records in the historical time, collect a sample telemetry accuracy set and a sample remote control accuracy set, and collect real point electrical quality parameters after checking and verification, and label to obtain a sample point checking quality parameter set; Based on AI machine learning, construct a point checking analysis network; Use the sample telemetry accuracy set, the sample remote control accuracy set and the sample point checking quality parameter set to supervise the training and testing of the point checking analysis network until the test accuracy is qualified; Combine the telemetry accuracy and the remote control accuracy of the same point in the telemetry accuracy matrix and the remote control accuracy matrix, input the point checking analysis network, output the basic point checking quality parameter, and obtain the basic point checking quality parameter matrix. 9.The AI recognition fusion-based substation site information checking method according to claim 7, characterized in that, According to the test reliability matrix, correct the basic point checking quality parameter matrix to obtain a point checking quality parameter matrix as a substation point information checking result, including: According to the test reliability matrix, configure a point checking quality correction coefficient matrix; The point position collation quality parameter matrix is corrected by using the point position collation quality correction coefficient matrix, and a point position collation quality parameter matrix is obtained as a substation point position information collation result.

10. A substation site information checking system based on AI recognition fusion, characterized in that, The system for implementing the AI recognition fusion-based substation point position information collation method of any one of claims 1-9 comprises: The first point position precision test module is configured to, after the hardware connection collation in the substation point position information collation is completed, select a first point position for remote measurement test and remote control test, and obtain a first remote measurement precision and a first remote control precision. The iterative test module is configured to select point positions electrically connected to the first point position according to a moving window, perform iterative remote measurement test and remote control test, and construct a remote measurement precision matrix and a remote control precision matrix. The precision verification module is configured to call historical point position collation data of point positions of the same family, perform precision verification on the remote measurement precision matrix and the remote control precision matrix, and obtain a test credibility matrix. The AI fusion recognition module is configured to perform AI fusion recognition according to the test credibility matrix, the remote measurement precision matrix, and the remote control precision matrix, obtain a point position collation quality parameter matrix, and use the point position collation quality parameter matrix as a substation point position information collation result.

Images