A test method and system for harmonic electrical fire monitoring

Abstract

The application discloses a test method and system for harmonic electrical fire monitoring, and relates to the technical field of electrical safety monitoring. A working condition benchmark determination module generates a current working condition identifier according to a test working condition record, and determines a working condition benchmark vector and a working condition boundary vector. A parameter deviation generation module compares a current parameter vector with the working condition benchmark vector and the working condition boundary vector, and generates a parameter deviation vector. A harmonic electrical correlation module pairs harmonic side parameters and electrical side parameters in the parameter deviation vector, and generates a correlation factor. A risk score generation module generates a risk score value and a risk state identifier according to the parameter deviation vector and the correlation factor. A test feedback revision module performs feedback revision according to the corresponding relationship between a test result identifier and the risk state identifier. The application can make the test evaluation process correspond to the load category, the operation stage and the test result, and is suitable for harmonic electrical fire monitoring test.

Description

Technical Field

[0001] This invention relates to the field of electrical safety monitoring technology, specifically to a test method and system for monitoring harmonic electrical fires. Background Technology

[0002] In harmonic electrical fire monitoring tests, existing methods typically use single parameter thresholds or several parameters independently as test criteria, failing to incorporate the relationship between load type, operating stage, and normal operation records into the same processing chain. Different loads exhibit different harmonic backgrounds during startup, stable operation, and switching. Using a unified benchmark for comparison can easily lead to normal load disturbances and electrical fire hazards being mixed into the same judgment scope. Furthermore, existing methods lack a fixed logical correlation between harmonic parameters and electrical parameters, and risk scoring boundaries and weights are often left to be manually assigned, making it difficult to revise test results in accordance with operational test records.

[0003] The aforementioned deficiencies mainly arise from the increase in nonlinear loads in electrical circuits, frequent switching of operating phases, and the gradual and intermittent nature of the formation of hidden dangers. When the contact state, insulation state, or local conduction state of the line changes, a single parameter may only show a local deviation and has not yet directly reached the alarm threshold; while during the start-up and shutdown of normal loads, short-term harmonic fluctuations or current changes may also occur.

[0004] Without a benchmark under the same operating conditions, parameter deviation vector, harmonic electrical correlation factor, and feedback revision process based on test result identification, there may be inconsistencies between the risk status and the actual test results, and subsequent tests may continue to use the original criteria, resulting in a disconnect between risk scoring, status mapping, and test records. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a test method and system for monitoring harmonic electrical fires, solving the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: a test system for monitoring harmonic electrical fires, comprising a working condition reference determination module, a parameter deviation generation module, a harmonic electrical correlation module, a risk score generation module, and a test feedback revision module; The operating condition reference determination module acquires the test operating condition records under normal electrical fire monitoring conditions, generates the current operating condition identifier Kt based on the test operating condition records, and determines the operating condition reference vector Bt and the operating condition boundary vector Et in the normal operation records. The parameter deviation generation module uses the working condition reference vector Bt and the working condition boundary vector Et as a basis, compares them with the current parameter vector Xt, and generates the parameter deviation vector Qst. The harmonic electrical correlation module correlates the parameters that deviate from the vector Qst and generates a correlation factor Mdt. The risk score generation module generates a basic score item FS based on the parameter deviation vector Qst, and uses the correlation factor Mdt to perform correlation correction on the basic score item to generate the risk score value RS and the risk status identifier YS. The test feedback revision module extracts the test result identifier Zs from the test condition record and provides feedback based on the correspondence between the risk status identifier YS and the test result identifier Zs.

[0007] Preferably, the operating condition reference determination module includes an operating condition identifier generation unit and an operating condition reference determination unit; The operating condition identifier generation unit obtains the test operating condition record corresponding to the t-th analysis window in the test system under normal electrical fire monitoring conditions, and extracts the load category identifier and operation phase identifier from the test operating condition record, and generates the current operating condition identifier Kt according to the combination relationship between the load category identifier and the operation phase identifier; The current operating condition identifier Kt is obtained by using the load category identifier as the first field, the running stage identifier as the second field, and generating the current operating condition identifier Kt in the form of a binary tuple. Among them, the test condition record comes from the condition record item formed during the test, the load category identifier indicates the load type corresponding to the t-th analysis window, and the operation stage identifier indicates the operation stage corresponding to the t-th analysis window; The same working condition benchmark determination unit uses the current working condition identifier Kt as the basis, and selects records that are the same as the current working condition identifier Kt from the existing normal operation records to form the same working condition normal operation record set Ωt; The formula for obtaining the normal operating record set Ωt under the same working conditions is: Ωt={r|Kr=Kt,Zr=N}; In the formula, Ωt specifically represents the set of normal operation records belonging to the same working condition as the t-th analysis window, r represents any historical analysis window record in the normal operation records, Kr represents the working condition identifier corresponding to the historical analysis window record r, Kt specifically represents the current working condition identifier corresponding to the t-th analysis window, Zr represents the test result identifier corresponding to the historical analysis window record r, and N represents the normal operation category; Among them, any historical analysis window record r in the normal operation record includes the following: total harmonic distortion rate Dr in historical analysis window record r, odd and even harmonic bias Pr in historical analysis window record r, harmonic phase offset Ar in historical analysis window record r, residual current fluctuation Lr in historical analysis window record r, load current change rate Cr in historical analysis window record r, and short-time impulse count Sr in historical analysis window record r; Based on the normal operating record set Ωt under the same operating conditions, the operating condition reference vector Bt = (bD, bP, bA, bL, bC, bS) is calculated, where bD represents the operating condition reference value of total harmonic distortion, bP represents the operating condition reference value of odd and even harmonic bias, bA represents the operating condition reference value of harmonic phase offset, bL represents the operating condition reference value of residual current fluctuation, bC represents the operating condition reference value of load current change rate, and bS represents the operating condition reference value of short-time impulse count. The formula is as follows: ; In the formula, bi represents the baseline value of parameter category i under the current operating condition identifier Kt, i represents the parameter category, and the value range is: total harmonic distortion rate D, odd and even harmonic offset P, harmonic phase offset A, residual current fluctuation L, load current change rate C, short-time impact count S, nt represents the number of records in the normal operation record set Ωt under the same operating condition, and x(r, i) represents the parameter value corresponding to parameter category i in the historical analysis window record r; Based on the normal operating record set Ωt and the operating condition reference vector Bt, the operating condition boundary vector Et = (eD, eP, eA, eL, eC, eS) is calculated, where eD represents the operating condition boundary value of total harmonic distortion, eP represents the operating condition boundary value of odd and even harmonic bias, eA represents the operating condition boundary value of harmonic phase offset, eL represents the operating condition boundary value of residual current fluctuation, eC represents the operating condition boundary value of load current change rate, and eS represents the operating condition boundary value of short-time impulse count. The formula is as follows: ; In the formula, ei represents the allowable offset scale of parameter category i under the current working condition identifier Kt, εi represents the calculation resolution correction amount corresponding to parameter category i, which comes from the calculation accuracy configuration of the parameter in the test system, γi represents the boundary conversion coefficient corresponding to parameter category i, which comes from the parameter configuration table in the test system, and max represents taking the larger value between the calculation result and the calculation resolution correction amount.

[0008] Preferably, the parameter deviation generation module includes a parameter alignment unit and a deviation amount generation unit; The parameter alignment unit receives the operating condition reference vector Bt and the operating condition boundary vector Et, and reads the data in the parameter vector Xt=(Dnt, Pnt, Ant, Lnt, Cnt, Snt) at the current time. It then establishes a one-to-one correspondence according to the parameter category and obtains the parameter alignment set HCt={(Dnt, bD, eD), (Pnt, bP, eP), (Ant, bA, eA), (Lnt, bL, eL), (Cnt, bC, eC), (Snt, bS, eS)}. In the formula, Dnt represents the total harmonic distortion rate at the current time, Pnt represents the odd and even harmonic bias at the current time, Ant represents the harmonic phase offset at the current time, Lnt represents the residual current fluctuation at the current time, Cnt represents the load current change rate at the current time, and Snt represents the short-time impulse count at the current time.

[0009] Preferably, the deviation generation unit calculates the deviation of the parameters relative to the corresponding working condition reference value according to the obtained parameter alignment set HCt, and generates the parameter deviation vector Qst; For parameter category i, the deviation is calculated as follows: ; In the formula, qi represents the parameter deviation of parameter category i, and xi represents the current parameter value corresponding to parameter category i in the parameter vector Xt at the current time; The parameter deviation vector Qst = (qD, qP, qA, qL, qC, qS); Wherein, qD represents the total harmonic distortion deviation, qP represents the odd and even harmonic bias deviation, qA represents the harmonic phase offset deviation, qL represents the residual current fluctuation deviation, qC represents the load current change rate deviation, and qS represents the short-time impulse count deviation. The deviation qi of parameter category i is analyzed to determine the degree of deviation, as follows: When qi=0, it means that the current parameter value of parameter category i is consistent with the corresponding working condition reference value; When 0 < qi < 1, it means that the current parameter value of parameter category i has an offset relative to the corresponding working condition reference value that does not reach the allowable offset scale. When qi=1, it means that the current parameter value of parameter category i is offset from the corresponding working condition reference value to the allowable offset scale.

[0010] Preferably, the harmonic electrical correlation module includes a deviation matching unit and a correlation factor generation unit; The deviation pairing unit forms a first associated pairing with the total harmonic distortion rate deviation qD and the residual current fluctuation deviation qL in the parameter deviation vector Qst, forms a second associated pairing with the odd and even harmonic bias deviation qP and the load current change rate deviation qC, and forms a third associated pairing with the harmonic phase offset deviation qA and the short-time impulse count deviation qS. The first, second, and third associated pairings form an associated pairing set Gdt={(qD, qL), (qP, qC), (qA, qS)}.

[0011] qD and qL pairing: Total harmonic distortion deviation qD represents the deviation of the overall distortion of the current or voltage waveform from the reference under the same operating condition; The deviation of residual current fluctuation qL represents the deviation of the change in residual current in the line from the baseline under the same operating conditions. The two are paired to characterize: when the overall waveform distortion and residual current variation occur simultaneously within the same analysis window, this window exhibits composite abnormal characteristics caused by insulation degradation, changes in leakage paths, or changes in abnormal conductive paths. qP and qC pairing: The deviation of odd and even harmonic offset qP represents the deviation of the structure relationship between odd and even harmonics relative to the reference under the same operating condition. The deviation of the load current change rate qC represents the deviation of the load current change process from the reference under the same operating conditions; The two are paired to characterize: when the harmonic structure change and the load current change process occur together in the same analysis window, the window has composite abnormal characteristics caused by load nonlinear change, contact state change or local conduction state change. qA and qS pairing: The harmonic phase offset deviation qA represents the deviation of the harmonic phase relationship relative to the reference under the same operating conditions; The short-term impact count deviation qS represents the deviation of the number of short-term impact events from the baseline under the same operating conditions; The two are paired to characterize: when changes in harmonic phase relationship and short-term impact events occur together in the same analysis window, the window has composite abnormal characteristics caused by transient disturbances, intermittent conduction, or contact discontinuity; Preferably, the association factor generation unit is based on the association pairing set Gdt, and calculates three association sub-items respectively: the first association sub-item m1, the second association sub-item m2 and the third association sub-item m3, with the following formulas: m1=qDqL, m2=qPqC, m3=qAqS; In the formula, m1 specifically represents the degree of common deviation between the total harmonic distortion rate deviation and the residual current fluctuation deviation, m2 represents the degree of common deviation between the odd and even harmonic bias deviation and the load current change rate deviation, and m3 represents the degree of common deviation between the harmonic phase offset deviation and the short-time impulse count deviation, and 0≤m1≤1, 0≤m2≤1, 0≤m3≤1. When any deviation in a pair is 0, the corresponding associated sub-item of that pair is 0; when both deviations in a pair are close to 1, the corresponding associated sub-item of that pair is close to 1; therefore, the associated sub-item is not the result of a single parameter deviation, but the result of the common deviation of two parameters within the same analysis window. Based on the three related sub-items, the correlation factor Mdt is calculated using the following formula: Mdt = a1m1 + a2m2 + a3m3; where a1 represents the first correlation coefficient, corresponding to the correlation pair (qD, qL); a2 represents the second correlation coefficient, corresponding to the correlation pair (qP, qC); a3 represents the third correlation coefficient, corresponding to the correlation pair (qA, qS), and 0 ≤ a1 ≤ 1, 0 ≤ a2 ≤ 1, 0 ≤ a3 ≤ 1, a1 + a2 + a3 = 1.

[0012] a1=N1 / (N1+N2+N3), a2=N2 / (N1+N2+N3), a2=N2 / (N1+N2+N3; In the formula, N1 represents the number of occurrences of the fault scenario category corresponding to the association pair (qD, qL), N2 represents the number of occurrences of the fault scenario category corresponding to the association pair (qP, qC), and N3 represents the number of occurrences of the fault scenario category corresponding to the association pair (qA, qS). Preferably, the risk score generation module includes a weighted score generation unit and a score boundary generation and state mapping unit; The weighted scoring generation unit reads the set of labeled test records for the same working condition corresponding to the current working condition identifier Kt; the set of labeled test records for the same working condition Φt={r|Or=Ot,Jr=Jt,Zr∈{N,H}}; In the formula, Or represents the load category identifier of the historical analysis window record r, Ot represents the load category identifier of the t-th analysis window, Jr represents the running stage identifier of the historical analysis window record r, Jt represents the running stage identifier of the t-th analysis window, N represents the normal operation category, and H represents the fault scenario category. The basic weight vector W = (wD, wP, wA, wL, wC, wS) is calculated based on the parameter separation degree between the normal operation category and the failure scenario category in the labeled test record set Φt under the same working conditions. The steps are as follows: S1. Let the set of parameter categories be I = {D, P, A, L, C, S}; S2. For any parameter category i∈I, first calculate the parameter deviation from the mean μ(i,N) under the normal operation category, and then calculate the parameter deviation from the mean μ(i,H) under the fault scenario category. The formula is as follows: ; ; In the formula, μ(i, N) specifically represents the parameter deviation mean of parameter category i in the normal operation category, μ(i, H) represents the parameter deviation mean of parameter category i in the fault scenario category, nN represents the number of records in the normal operation category in the set of labeled test records Φt under the same working condition, nH represents the number of records in the set of labeled test records Φt under the same working condition, and q(r, i) represents the parameter deviation amount corresponding to parameter category i in the historical analysis window record r; S3. Based on the parameter deviation mean μ(i, N) under the normal operation category and the parameter deviation mean μ(i, H) under the fault scenario category, calculate the deviation discreteness d(i, N) under the normal operation category and the deviation discreteness d(i, H) under the fault scenario category; the formulas are as follows: ; ; S4. First, calculate the class separation degree ηi of parameter class i, then normalize the class separation degree of each parameter class to obtain the basic weight wi, and obtain the basic weight vector W based on the basic weight wi. The formula is as follows: ; In the formula, εw represents the correction amount of the denominator in the weight calculation, which takes the value of 0.001 and is derived from the scoring calculation accuracy configuration of the test system; ; In the formula, j represents any parameter category used for summation in the parameter category set; And 0≤wi≤1, wD+wP+wA+wL+wC+wS=1; Based on the basic weight vector W, and combined with the current parameter deviation vector Qst, a basic scoring item FS is generated; The basic scoring item FS is obtained using the following formula: ; Expanded to: FS = wDqD + wPqP + wAqA + wLqL + wCqC + wSqS; Since the deviation of each parameter qiq ranges from 0 to 1, and the sum of each basic weight is 1, therefore: 0 ≤ FS ≤ 1; The risk score value RS is calculated by combining the basic scoring item FS with the correlation factor Mdt, as shown in the following formula; the range of the risk score value RS is: 0≤RS≤100; ; In the formula, min means that when the result after correlation correction exceeds 1, it is counted as 1; 100 means that the score result is converted into a score range of 0 to 100; and λ represents the correlation correction coefficient, which ranges from 0 to 1.

[0013] The correlation correction coefficient λ is obtained as follows: ; ; ; In the formula, pMN represents the mean of the correlation factors in the normal operation category, pMH represents the mean of the correlation factors in the fault scenario category, and Mdr represents the correlation factor corresponding to record r in the historical analysis window. Preferably, the scoring boundary generation and state mapping unit compares the risk score value RS with the set first scoring boundary θ1 and second scoring boundary θ2 to obtain the risk state identifier YS; Risk status identifier YS is obtained through matching in the following way: When 0 ≤ RS < θ1, YS = Y1; When θ1≤RS<θ2, YS=Y2; When θ2≤RS≤100, YS=Y3; In the formula, Y1 represents the first risk state, Y2 represents the second risk state, and Y3 represents the third risk state.

[0014] The first scoring boundary θ1 and the second scoring boundary θ2 are obtained in the following way: S1. For any historical analysis window record r in the set of marked test records Φt under the same working condition, calculate the historical basic score item F(S, r) according to the following formula. ; ; In the formula, F(S, r) specifically represents the basic scoring item corresponding to record r in the historical analysis window, q(r, i) represents the parameter deviation corresponding to parameter category i in record r in the historical analysis window, and R(S, r) represents the risk score value corresponding to record r in the historical analysis window. S2. Calculate the score center CN for the normal operation category and the score center CH for the failure scenario category; ; ; S3. Under the condition that CN < CH, the first scoring boundary θ1 and the second scoring boundary θ2 are generated according to the following formula: θ1=CN+(CH-CN) / 3; θ2=CN+2×(CH-CN) / 3; and 0≤θ1<θ2≤100; Preferably, the test feedback revision module includes a feedback relationship discrimination unit and a parameter revision feedback unit; The feedback relationship discrimination unit extracts the test result identifier Zs from the test condition record corresponding to the t-th analysis window, and maps it to the risk status identifier YS to generate the baseline revision mark cB and the score revision direction gs; Where Zs∈{N,H}; The benchmark revision flag cB is used to control whether the current parameter vector can enter the revision process of the operating condition benchmark vector Bt and the operating condition boundary vector E; ; In the formula, Zs=N indicates that the test result of the t-th analysis window is identified as the normal operation category, and Zs=H indicates that the test result of the t-th analysis window is identified as the fault scenario category; The score revision direction (gs) is used to control the revision direction of score weights and correlation factor correction coefficients; ; In the formula, gs=1 indicates that the failure scenario category is not mapped to the third risk state, and the deviation and correlation corresponding to this analysis window participate in the positive revision; gs=-1 indicates that the normal operation category is mapped to the second or third risk state, and the deviation and correlation corresponding to this analysis window participate in the reverse revision; gs=0 indicates that the correspondence between the risk state identifier YS and the test result identifier Zs is consistent, and the scoring weight and correlation factor correction coefficient remain at their current values. The parameter revision feedback unit revises the working condition baseline vector Bt, working condition boundary vector Et, scoring weight wi, and correlation factor correction coefficient λ according to the baseline revision mark cB and the scoring revision direction gs, and then feeds it back. The revision method for the operating condition reference vector Bt is as follows: ; In the formula, nbi represents the revised working condition baseline value of parameter category i, and ρB represents the working condition baseline revision coefficient, with a value range of 0 < ρB ≤ 0.2; The result of this formula means that when cB=1, the parameter value xi of the current normal operation category participates in the revision of the operating condition baseline value; when cB=0, nbi=bi, and the parameter value corresponding to the fault scenario category does not enter the operating condition baseline vector. The method for revising the working condition boundary vector Et is as follows: ; In the formula, nei represents the revised working condition boundary value of parameter category i, and ρE represents the working condition boundary revision coefficient, with a value range of 0 < ρE ≤ 0.2; The revision method for the scoring weight wi is as follows: ; In the formula, nwi represents the intermediate weight of parameter category i, ρW represents the weight revision coefficient, with a value range of 0 < ρW ≤ 0.1, and j represents any parameter category in the parameter category set; The intermediate weights of parameter category i are normalized to obtain the revised score weight znwi, as shown in the following formula: ; The revision method for the correlation factor correction coefficient λ is as follows: ; In the formula, ρλ represents the revision coefficient of the correlation factor correction coefficient, and its value ranges from 0 to 0.1.

[0015] The result of this formula means: when the failure scenario category is not mapped to the third risk state, gs=1, and the correlation factor Mdt participates in the positive revision; when the normal operation category is mapped to the second or third risk state, gs=-1, and the correlation factor Mdt participates in the negative revision; when gs=0, nλ=λ. A test method for monitoring harmonic electrical fires includes the following steps: Step 1: The working condition reference determination module acquires the test working condition record under normal electrical fire monitoring status, generates the current working condition identifier Kt based on the test working condition record, and determines the working condition reference vector Bt and the working condition boundary vector Et in the normal operation record. Step 2: The parameter deviation generation module uses the working condition reference vector Bt and the working condition boundary vector Et as a basis, compares them with the current parameter vector Xt, and generates the parameter deviation vector Qst. Step 3: The harmonic electrical correlation module correlates the parameters inside the parameter deviation vector Qst to generate the correlation factor Mdt; Step 4: The risk score generation module generates a basic score item FS based on the parameter deviation vector Qst, and uses the correlation factor Mdt to perform correlation correction on the basic score item to generate the risk score value RS and the risk status identifier YS. Step 5: The test feedback revision module extracts the test result identifier Zs from the test condition record and provides feedback based on the correspondence between the risk status identifier YS and the test result identifier Zs.

[0016] This invention provides a test method and system for monitoring harmonic electrical fires, which has the following beneficial effects: (1) This invention, through continuous processing of operating condition benchmark determination, parameter deviation generation, harmonic electrical correlation, risk score generation, and test feedback revision, enables harmonic electrical fire monitoring tests to move beyond simple parameter comparisons or fixed threshold judgments. Instead, it establishes a complete processing chain around test conditions, normal operation records, current parameter changes, and test result feedback. This approach allows parameter judgments under different load categories and different operating stages to have corresponding operating condition references, improving the test evaluation bias caused by using the same judgment benchmark for different operating conditions.

[0017] (2) By comparing the current parameters with the benchmark and boundary of the same operating condition item by item, this invention can convert the total harmonic distortion rate, odd and even harmonic offset, harmonic phase offset, residual current fluctuation, load current change rate, and short-time impact count into deviation results with a unified meaning. This method allows parameters with different dimensions and different numerical ranges to enter the same judgment process, which is beneficial for distinguishing between normal operation fluctuations and parameter deviations related to electrical fire hazards.

[0018] (3) This invention pairs harmonic deviation results with electrical deviation results and forms a correlation judgment based on the degree of common deviation, so that risk assessment no longer depends on the isolated change of a single parameter. This method can reflect the combined relationship between waveform distortion, residual current change, load current change and short-term impact within the same analysis window, which is beneficial for identifying composite abnormal characteristics under operating conditions such as insulation degradation, contact state change and intermittent conduction.

[0019] (4) This invention combines risk scoring and test feedback revision, enabling test results to have a reverse effect on subsequent operating condition benchmarks, operating condition boundaries, scoring weights, and related correction parameters. Data in the normal operation category can be included in the normal reference system, while data in the failure scenario category is not included in the normal reference range; when the risk status is inconsistent with the test results, the scoring rules can be adjusted according to the feedback direction. Attached Figure Description

[0020] Figure 1 This is a schematic flowchart of a test system for monitoring harmonic electrical fires according to the present invention. Figure 2 This is a schematic diagram of the test method steps for monitoring harmonic electrical fires according to the present invention; Figure 3 This is a schematic diagram of the scoring revision direction determination process of the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0022] Example 1 This invention provides a test system for monitoring harmonic electrical fires. Please refer to [link / reference]. Figure 1 It includes a working condition baseline determination module, a parameter deviation generation module, a harmonic electrical correlation module, a risk score generation module, and a test feedback revision module; The operating condition reference determination module acquires the test operating condition records under normal electrical fire monitoring conditions, generates the current operating condition identifier Kt based on the test operating condition records, and determines the operating condition reference vector Bt and the operating condition boundary vector Et in the normal operation records. The parameter deviation generation module uses the working condition reference vector Bt and the working condition boundary vector Et as a basis, compares them with the current parameter vector Xt, and generates the parameter deviation vector Qst. The harmonic electrical correlation module correlates the parameters that deviate from the vector Qst and generates a correlation factor Mdt. The risk score generation module generates a basic score item FS based on the parameter deviation vector Qst, and uses the correlation factor Mdt to perform correlation correction on the basic score item to generate the risk score value RS and the risk status identifier YS. The test feedback revision module extracts the test result identifier Zs from the test condition record and provides feedback based on the correspondence between the risk status identifier YS and the test result identifier Zs.

[0023] In this embodiment, the operating condition reference determination module generates a current operating condition identifier based on the test operating condition record, and determines the operating condition reference vector and operating condition boundary vector in the normal operation record. This processing method allows subsequent judgments to be based on the normal operating state corresponding to the load category and operating stage, instead of using a single fixed reference, and is applicable to harmonic electrical fire monitoring tests under different loads and different operating stages.

[0024] The parameter deviation generation module compares the current parameter vector with the operating condition reference vector and the operating condition boundary vector to generate a parameter deviation vector Qst. Through this processing, the total harmonic distortion rate, odd and even harmonic offset, harmonic phase offset, residual current fluctuation, load current change rate, and short-time impulse count can participate in subsequent judgments under the same deviation scale, solving the problem of different parameters having different dimensions and the difficulty of direct comparison.

[0025] The harmonic electrical correlation module correlates the internal parameters of the parameter deviation vector Qst to generate a correlation factor Mdt. This process does not judge each parameter in isolation, but combines harmonic changes with electrical changes such as residual current, load current changes, and short-term impacts for analysis, so that risk assessment can reflect the common changes of multiple anomalies within the same analysis window.

[0026] The risk score generation module first generates a basic score item FS based on the parameter deviation vector Qst, and then uses the correlation factor Mdt to correct the basic score item FS to obtain the risk score value RS and the risk status indicator YS. This structure allows the risk score to reflect both the deviation of individual parameters and the correlation between parameters, which is different from the judgment method that relies solely on a single threshold or the simple addition of parameters.

[0027] The test feedback revision module extracts the test result identifier Zs from the test condition record and provides feedback based on the correspondence between the risk status identifier YS and the test result identifier Zs. This processing allows the operating condition baseline, operating condition boundary, scoring weight, correlation coefficient, and correlation factor correction coefficient to be revised along with the test record, ensuring that subsequent analysis windows use the processed parameters corrected by the test results. Because this embodiment first determines the operating condition baseline vector Bt and operating condition boundary vector Et under the same operating condition based on the current operating condition identifier Kt, and then performs parameter deviation and correlation analysis, normal fluctuations caused by normal load start-up and shutdown, operation phase switching, etc., will not directly enter the risk status judgment according to a uniform threshold, which is more in line with the actual operating scenario in harmonic electrical fire monitoring tests.

[0028] Example 2 This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically: the working condition reference determination module includes a working condition identification generation unit and a working condition reference determination unit; The operating condition identifier generation unit obtains the test operating condition record corresponding to the t-th analysis window in the test system under normal electrical fire monitoring conditions, and extracts the load category identifier and operation phase identifier from the test operating condition record, and generates the current operating condition identifier Kt according to the combination relationship between the load category identifier and the operation phase identifier; The current operating condition identifier Kt is obtained by using the load category identifier as the first field, the running stage identifier as the second field, and generating the current operating condition identifier Kt in the form of a binary tuple. The same working condition benchmark determination unit uses the current working condition identifier Kt as the basis, and selects records that are the same as the current working condition identifier Kt from the existing normal operation records to form the same working condition normal operation record set Ωt; The formula for obtaining the normal operating record set Ωt under the same working conditions is: Ωt={r|Kr=Kt,Zr=N}; In the formula, Ωt specifically represents the set of normal operation records belonging to the same working condition as the t-th analysis window, r represents any historical analysis window record in the normal operation records, Kr represents the working condition identifier corresponding to the historical analysis window record r, Kt specifically represents the current working condition identifier corresponding to the t-th analysis window, Zr represents the test result identifier corresponding to the historical analysis window record r, and N represents the normal operation category; Among them, any historical analysis window record r in the normal operation record includes the following: total harmonic distortion rate Dr in historical analysis window record r, odd and even harmonic bias Pr in historical analysis window record r, harmonic phase offset Ar in historical analysis window record r, residual current fluctuation Lr in historical analysis window record r, load current change rate Cr in historical analysis window record r, and short-time impulse count Sr in historical analysis window record r; Based on the set of normal operating records Ωt under the same working conditions, the working condition reference vector Bt = (bD, bP, bA, bL, bC, bS) is calculated as follows: ; In the formula, bi represents the baseline value of parameter category i under the current operating condition identifier Kt, i represents the parameter category, and the value range is: total harmonic distortion rate D, odd and even harmonic offset P, harmonic phase offset A, residual current fluctuation L, load current change rate C, short-time impact count S, nt represents the number of records in the normal operation record set Ωt under the same operating condition, and x(r, i) represents the parameter value corresponding to parameter category i in the historical analysis window record r; Based on the normal operating record set Ωt and the operating condition reference vector Bt, the operating condition boundary vector Et = (eD, eP, eA, eL, eC, eS) is calculated as follows: ; In the formula, ei represents the allowable offset scale of parameter category i under the current working condition identifier Kt, εi represents the calculation resolution correction amount corresponding to parameter category i, γi represents the boundary conversion coefficient corresponding to parameter category i, and max represents taking the larger value between the calculation result and the calculation resolution correction amount.

[0029] The steps for obtaining the computational resolution correction εi corresponding to parameter category i are as follows: Read the output accuracy configuration of each parameter in the experimental system; Let Δi be the minimum output resolution corresponding to parameter category i. The minimum output resolution Δi is used as the resolution correction factor εi.

[0030] The steps to obtain the boundary conversion coefficients corresponding to parameter category i are as follows: The operating condition baseline determination module filters historical analysis window records that are the same as the current operating condition identifier Kt and whose test results are identified as normal operation category, based on the current operating condition identifier Kt, to form a set of normal operation records under the same operating condition Ωt. Calculate the baseline value bi of parameter category i in the normal operating record set Ωt under the same operating conditions; Calculate the offset of each historical parameter value of parameter category i relative to the operating condition baseline value bi; The average value of the offsets is taken as the average offset, and the maximum value among the offsets is taken as the boundary offset; The ratio of the boundary offset to the average offset is used as the boundary conversion factor γi; the value range is 1≤γi≤5; In this embodiment, the operating condition reference determination module generates a current operating condition identifier Kt by combining the load category identifier and the operating stage identifier in a binary form. Using the current operating condition identifier Kt as a filtering condition, it forms a set of normal operating records Ωt from the normal operating records. This processing method ensures that the judgment basis for the t-th analysis window no longer comes from a unified fixed reference, but rather from normal operating records that match the current load category and current operating stage. Therefore, harmonic backgrounds generated by variable frequency loads, resistive loads, and switching power supply loads at different operating stages can each form corresponding references, effectively solving the problem of using the same judgment reference for different load conditions in existing harmonic electrical fire monitoring tests.

[0031] In this embodiment, the same-condition reference determination unit calculates the operating condition reference vector Bt based on the normal operating condition record set Ωt, so that the total harmonic distortion rate, odd and even harmonic offset, harmonic phase offset, residual current fluctuation, load current change rate, and short-time impulse count all have corresponding normal operating reference values. This structure can compare the current parameters with historical normal parameters under the same operating condition item by item, rather than judging solely based on the current instantaneous values. This is beneficial for distinguishing between operating condition fluctuations during normal operation and parameter offsets related to fire hazards, providing the subsequent parameter deviation generation module with a comparison basis with a clear source.

[0032] In this embodiment, the boundary vector Et is not arbitrarily given, but is jointly determined by the calculated resolution correction εi, the boundary conversion coefficient γi, and the parameter offsets in the normal operation records under the same operating conditions. εi originates from the output accuracy configuration of each parameter in the test system, and γi originates from the ratio of the boundary offset to the average offset in the normal operation records under the same operating conditions. This approach ensures that the allowable offset scale is constrained by both the equipment output accuracy and the normal fluctuation range under the same operating conditions. It prevents parameter deviations and calculation distortions caused by excessively small denominator values, and also prevents inconsistencies in scale when different parameters with different dimensions, resolutions, or normal fluctuation ranges are included in the same comparison process.

[0033] In this embodiment, the current operating condition identifier Kt, the operating condition reference vector Bt, and the operating condition boundary vector Et together constitute the prerequisites for subsequent parameter deviation calculations. This enables the parameter deviation generation module to calculate according to the "same operating condition reference value" and the "same operating condition allowable offset scale" when processing the current parameter vector Xt. This setting, in conjunction with subsequent harmonic electrical association, risk scoring, and test feedback revision processes, ensures that the risk judgment link of the test system has a unified data source and transmission relationship. This helps to improve the problems in existing tests where risk status is disconnected from actual test results, and normal load disturbances and hidden danger characteristics are mixed in the same judgment range.

[0034] Example 3 This embodiment is an explanation based on Embodiment 2. Please refer to it. Figure 1 Specifically: the parameter deviation generation module includes a parameter alignment unit and a deviation amount generation unit; The parameter alignment unit receives the operating condition reference vector Bt and the operating condition boundary vector Et, and reads the data in the parameter vector Xt=(Dnt, Pnt, Ant, Lnt, Cnt, Snt) at the current time. It then establishes a one-to-one correspondence according to the parameter category and obtains the parameter alignment set HCt={(Dnt, bD, eD), (Pnt, bP, eP), (Ant, bA, eA), (Lnt, bL, eL), (Cnt, bC, eC), (Snt, bS, eS)}. In the formula, Dnt represents the total harmonic distortion rate at the current time, Pnt represents the odd and even harmonic bias at the current time, Ant represents the harmonic phase offset at the current time, Lnt represents the residual current fluctuation at the current time, Cnt represents the load current change rate at the current time, and Snt represents the short-time impulse count at the current time.

[0035] The deviation generation unit calculates the deviation of the parameters relative to the corresponding working condition reference value based on the obtained parameter alignment set HCt, and generates the parameter deviation vector Qst=(qD, qP, qA, qL, qC, qS). For parameter category i, the deviation is calculated as follows: ; In the formula, qi represents the parameter deviation of parameter category i, and xi represents the current parameter value corresponding to parameter category i in the parameter vector Xt at the current time; Wherein, qD represents the total harmonic distortion deviation, qP represents the odd and even harmonic bias deviation, qA represents the harmonic phase offset deviation, qL represents the residual current fluctuation deviation, qC represents the load current change rate deviation, and qS represents the short-time impulse count deviation. The deviation qi of parameter category i is analyzed to determine the degree of deviation, as follows: When qi=0, it means that the current parameter value of parameter category i is consistent with the corresponding working condition reference value; When 0 < qi < 1, it means that the current parameter value of parameter category i has an offset relative to the corresponding working condition reference value that does not reach the allowable offset scale. When qi=1, it means that the current parameter value of parameter category i is offset from the corresponding working condition reference value to the allowable offset scale.

[0036] In this embodiment, the parameter deviation generation module, based on the operating condition reference vector Bt and the operating condition boundary vector Et, performs item-by-item alignment of the current time parameter vector Xt to form a parameter alignment set HCt. This processing method binds the current parameter, the reference value under the same operating condition, and the allowable offset scale under the same operating condition one by one according to the parameter category, avoiding confusion in the correspondence of total harmonic distortion rate, odd and even harmonic offset, harmonic phase offset, residual current fluctuation, load current change rate, and short-time impact count in subsequent calculations, and making the parameter comparison in the current analysis window have a clear data source and calculation object.

[0037] In this embodiment, the deviation generation unit does not directly use the current parameter values ​​for risk assessment. Instead, it first calculates the deviation of each parameter relative to the corresponding operating condition baseline value based on the parameter alignment set HCt, and generates a parameter deviation vector Qst. This processing method converts parameters with different dimensions, numerical ranges, and output precisions into parameter deviations with a unified value range, enabling different types of parameters such as total harmonic distortion rate, residual current fluctuation, and load current change rate to enter the subsequent harmonic electrical correlation and risk scoring process on the same scale.

[0038] In this embodiment, the parameter deviation is calculated based on the current parameter value, the operating condition reference value, and the operating condition boundary value, reflecting the degree of deviation of the current parameter relative to the normal operating state under the same operating condition. When qi=0, it indicates that the current parameter value is consistent with the corresponding operating condition reference value; when 0<qi<1, it indicates that the current parameter value is within the deviation range of the operating condition boundary value; when qi=1, it indicates that the deviation of the current parameter value reaches or exceeds the corresponding operating condition boundary value. Thus, the test system can transform the original parameter changes into deviation results with a unified meaning, facilitating subsequent modules to identify the source and extent of parameter changes.

[0039] In this embodiment, the parameter alignment unit and the deviation generation unit are sequentially connected, so that the parameter deviation vector Qst retains the difference between the current parameter and the normal operating state under the same working condition, while eliminating the impact of inconsistent dimensions between different parameters on subsequent processing. This parameter deviation vector serves as the common input of the harmonic electrical correlation module and the risk scoring generation module, ensuring that the subsequent correlation factor Mdt, basic scoring item FS, risk score value RS, and risk status identifier YS are all based on the same deviation scale. This helps to improve the problems in existing test processing, such as the difficulty in judging single parameter thresholds, direct comparison of different parameters, and the difficulty in distinguishing between normal load disturbances and hidden danger characteristics.

[0040] Example 4 This embodiment is an explanation based on Embodiment 3. Please refer to it. Figure 1 Specifically: the harmonic electrical correlation module includes a deviation matching unit and a correlation factor generation unit; The deviation pairing unit forms a first associated pairing with the total harmonic distortion rate deviation qD and the residual current fluctuation deviation qL in the parameter deviation vector Qst, forms a second associated pairing with the odd and even harmonic bias deviation qP and the load current change rate deviation qC, and forms a third associated pairing with the harmonic phase offset deviation qA and the short-time impulse count deviation qS. The first, second, and third associated pairings form an associated pairing set Gdt={(qD, qL), (qP, qC), (qA, qS)}.

[0041] The association factor generation unit is based on the association pairing set Gdt, and calculates three association sub-items respectively: the first association sub-item m1, the second association sub-item m2 and the third association sub-item m3, with the following formulas: m1=qDqL, m2=qPqC, m3=qAqS; In the formula, m1 specifically represents the degree of common deviation between the total harmonic distortion rate deviation and the residual current fluctuation deviation, m2 represents the degree of common deviation between the odd and even harmonic bias deviation and the load current change rate deviation, and m3 represents the degree of common deviation between the harmonic phase offset deviation and the short-time impulse count deviation, and 0≤m1≤1, 0≤m2≤1, 0≤m3≤1. Based on the three related sub-items, the correlation factor Mdt is calculated using the following formula: Mdt = a1m1 + a2m2 + a3m3; In the formula, a1 represents the first correlation coefficient, corresponding to the correlation pair (qD, qL); a2 represents the second correlation coefficient, corresponding to the correlation pair (qP, qC); a3 represents the third correlation coefficient, corresponding to the correlation pair (qA, qS), and 0≤a1≤1, 0≤a2≤1, 0≤a3≤1, a1+a2+a3=1.

[0042] The risk scoring generation module includes a weighted scoring generation unit and a scoring boundary generation and state mapping unit; The weighted scoring generation unit reads the set of labeled test records for the same working condition corresponding to the current working condition identifier Kt; the set of labeled test records for the same working condition Φt={r|Or=Ot,Jr=Jt,Zr∈{N,H}}; In the formula, Or represents the load category identifier of the historical analysis window record r, Ot represents the load category identifier of the t-th analysis window, Jr represents the running stage identifier of the historical analysis window record r, Jt represents the running stage identifier of the t-th analysis window, N represents the normal operation category, and H represents the fault scenario category. The basic weight vector W = (wD, wP, wA, wL, wC, wS) is calculated based on the parameter separation degree between the normal operation category and the failure scenario category in the labeled test record set Φt under the same working conditions. The steps are as follows: S1. Let the set of parameter categories be I = {D, P, A, L, C, S}; S2. For any parameter category i∈I, first calculate the parameter deviation from the mean μ(i,N) under the normal operation category, and then calculate the parameter deviation from the mean μ(i,H) under the fault scenario category. S3. Based on the parameter deviation mean μ(i, N) under the normal operation category and the parameter deviation mean μ(i, H) under the fault scenario category, calculate and obtain the deviation discrete d(i, N) under the normal operation category and the deviation discrete d(i, H) under the fault scenario category. S4. First, calculate the class separation degree ηi of parameter class i, then normalize the class separation degree of each parameter class to obtain the basic weight wi, and obtain the basic weight vector W based on the basic weight wi. Based on the basic weight vector W, and combined with the current parameter deviation vector Qst, a basic scoring item FS is generated; The basic scoring item FS is obtained using the following formula: ; The risk score value RS is calculated by combining the basic scoring item FS with the correlation factor Mdt, as shown in the following formula; ; In the formula, min means that when the result after correlation correction exceeds 1, it is counted as 1; 100 means that the score result is converted into a score range of 0 to 100; and λ means the correlation correction coefficient.

[0043] The scoring boundary generation and state mapping unit compares the risk score value RS with the set first scoring boundary θ1 and second scoring boundary θ2 to obtain the risk state identifier YS. Risk status identifier YS is obtained through matching in the following way: When 0 ≤ RS < θ1, YS = Y1; When θ1≤RS<θ2, YS=Y2; When θ2≤RS≤100, YS=Y3; In the formula, Y1 represents the first risk state, Y2 represents the second risk state, and Y3 represents the third risk state.

[0044] In this embodiment, the harmonic electrical correlation module pairs the harmonic-side deviation and the electrical-side deviation in the parameter deviation vector Qst to form a correlation pairing set Gd. Specifically, the total harmonic distortion rate deviation qD and the residual current fluctuation deviation qL form the first correlation pairing; the odd / even harmonic bias deviation qP and the load current change rate deviation qC form the second correlation pairing; and the harmonic phase offset deviation qA and the short-time impulse count deviation qS form the third correlation pairing. This processing method ensures that harmonic variations are no longer treated as isolated parameters in the judgment, but rather participate in the analysis together with parameters that reflect changes in the line's electrical state. This helps to improve the problem in existing technologies where it is difficult to distinguish between normal load disturbances and hidden danger characteristics when relying solely on harmonic parameters or solely on electrical parameters for judgment.

[0045] In this embodiment, the correlation factor generation unit calculates the first correlation sub-item m1=qDqL, the second correlation sub-item m2=qPqC, and the third correlation sub-item m3=qAqS based on the correlation pairing set Gdt, and generates the correlation factor Mdt based on the first correlation coefficient a1, the second correlation coefficient a2, and the third correlation coefficient a3. Since each correlation sub-item is obtained by multiplying a harmonic deviation and an electrical deviation, when either deviation is absent, the corresponding correlation sub-item is not included in the common deviation result; when both deviations appear simultaneously within the same analysis window, the corresponding correlation sub-item is included in the correlation factor. Therefore, the correlation factor Mdt can characterize the common deviation relationship between harmonic characteristic changes and electrical operation changes within the same analysis window, enabling subsequent risk scoring to be corrected based on combined characteristics, rather than solely based on single parameter offsets.

[0046] In this embodiment, the risk scoring generation module calculates the basic weight vector W based on the set of labeled test records Φt corresponding to the current operating condition identifier Kt. The basic weight wi is obtained from the degree of parameter separation between the normal operation category and the fault scenario category, ensuring that the participation ratio of different parameters in the basic scoring item FS comes from the labeled test records under the same operating condition, rather than being manually fixed. This processing method enables total harmonic distortion, odd and even harmonic bias, harmonic phase shift, residual current fluctuation, load current change rate, and short-time impulse count to participate in the scoring according to their distinguishing ability in the same operating condition test, improving the problem of the disconnect between the weights of various parameters and the actual operating condition records in existing test evaluations.

[0047] In this embodiment, the risk score generation module first generates a basic score item FS based on the basic weight vector WWW and the parameter deviation vector Qst. Then, it combines the basic score item FS with the correlation factor Mdt to generate a risk score value RS, and maps it to the risk status identifier YS according to the first scoring boundary θ1 and the second scoring boundary θ2. This processing link ensures that the risk status identifier is simultaneously constrained by the degree of parameter deviation, parameter distinguishing ability, and the common deviation relationship of harmonic electrical components. This allows single-parameter fluctuations and multi-parameter anomalies to have different impacts on the scoring results. Therefore, the test system can generate risk assessment results that better fit the test records under different load categories and operating stages, and provide a clear scoring basis for the subsequent test feedback revision module to provide feedback based on the test result identifier Zs.

[0048] Example 5 This embodiment is an explanation based on Embodiment 4. Please refer to it. Figure 3 Specifically: the test feedback revision module includes a feedback relationship discrimination unit and a parameter revision feedback unit; The feedback relationship discrimination unit extracts the test result identifier Zs from the test condition record corresponding to the t-th analysis window, and maps it to the risk status identifier YS to generate the baseline revision mark cB and the score revision direction gs; Where Zs∈{N,H}; The benchmark revision flag cB is used to control whether the current parameter vector can enter the revision process of the operating condition benchmark vector Bt and the operating condition boundary vector E; ; In the formula, Zs=N indicates that the test result of the t-th analysis window is identified as the normal operation category, and Zs=H indicates that the test result of the t-th analysis window is identified as the fault scenario category; The score revision direction (gs) is used to control the revision direction of score weights and correlation factor correction coefficients; ; In the formula, gs=1 indicates that the fault scenario category is not mapped to the third risk state, gs=-1 indicates that the normal operation category is mapped to the second or third risk state, and gs=0 indicates that the correspondence between the risk state identifier YS and the test result identifier Zs is consistent. The parameter revision feedback unit revises the working condition baseline vector Bt, working condition boundary vector Et, scoring weight wi, and correlation factor correction coefficient λ according to the baseline revision mark cB and the scoring revision direction gs, and then feeds it back. The revision method for the operating condition reference vector Bt is as follows: ; In the formula, nbi represents the revised working condition baseline value of parameter category i, and ρB represents the working condition baseline revision coefficient; The method for revising the working condition boundary vector Et is as follows: ; In the formula, nei represents the revised boundary value of the working condition for parameter category i, and ρE represents the boundary revision coefficient for the working condition. The revision method for the scoring weight wi is as follows: ; In the formula, nwi represents the intermediate weight of parameter category i, ρW represents the weight revision coefficient, with a value range of 0 < ρW ≤ 0.1, and j represents any parameter category in the parameter category set; The intermediate weights of parameter category i are normalized to obtain the revised score weight znwi, as shown in the following formula: ; The revision method for the correlation factor correction coefficient λ is as follows: ; In the formula, ρλ represents the revision coefficient of the correlation factor correction coefficient, and its value ranges from 0 to 0.1.

[0049] In this embodiment, the test feedback revision module extracts the test result identifier Zs from the test condition record corresponding to the t-th analysis window through the feedback relationship discrimination unit, and maps the test result identifier Zs to the risk status identifier YS to generate a baseline revision marker cB and a score revision direction gs. The baseline revision marker cB is used to distinguish whether the current parameter vector can enter the revision process of the condition baseline vector Bt and the condition boundary vector Et, and the score revision direction gs is used to distinguish the revision direction of the score weight wi and the correlation factor correction coefficient λ. This processing method ensures that the test result identifier is not only saved as a test record, but also directly participates in the adjustment of the judgment parameters in the next analysis window, solving the problem of the disconnect between the risk status output and the test result record in the existing test system.

[0050] In this embodiment, when the test result identifier Zs=N, the benchmark revision flag cB=1, and the current parameter vector can participate in the revision of the operating condition benchmark vector Bt and the operating condition boundary vector Et; when the test result identifier Zs=HZ_s=HZs=H, the benchmark revision flag cB=0, and the current parameters under the fault scenario do not enter the normal operating condition benchmark and normal operating condition boundary. Through the calculation method of nbi and nei, parameter changes under the normal operation category can enter the normal reference system of the same operating condition, while parameter changes under the fault scenario category will not be written into the normal benchmark. Therefore, the operating condition benchmark and operating condition boundary can continuously correspond to the normal operation category, which is suitable for solving the problem of normal load disturbance and fault scenario parameters being mixed in the same benchmark range.

[0051] In this embodiment, the score revision direction gs is determined based on the correspondence between the test result identifier Zs and the risk state identifier YS. When the fault scenario category is not mapped to the third risk state, gs=1, indicating that the current deviation item needs to have a larger participation ratio in subsequent scoring; when the normal operation category is mapped to the second or third risk state, gs=-1, indicating that the current deviation item needs to have a smaller participation ratio in subsequent scoring; when the correspondence between the two is consistent, gs=0, and the scoring-related parameters remain in their current state. This processing method makes the revision of the score weight wi no longer dependent on manual judgment, but is triggered by the correspondence deviation between the test result and the risk state, allowing the score weight to be adjusted around the actual test result.

[0052] In this embodiment, the parameter revision feedback unit calculates the intermediate weights using nwi and normalizes them to obtain the revised scoring weights znwi; simultaneously, it revises the correlation factor correction coefficient λ using nλ. This processing method ensures that the parameter deviation vector Qst and correlation factor Mdt in the current analysis window can enter the subsequent scoring structure according to the direction of gs, so that the risk score value RS in the next analysis window is simultaneously constrained by the test results, the degree of parameter deviation, and the degree of harmonic electrical correlation. Thus, the test system forms a closed-loop processing link of "risk status output - test result verification - parameter revision feedback - next window judgment", which is suitable for solving the problem that test evaluation rules are difficult to change with the recorded operating conditions under fixed threshold and fixed weight conditions.

[0053] Example 6 A test method for monitoring harmonic electrical fires; please refer to [reference needed]. Figure 2 Specifically, it includes the following steps: Step 1: The working condition reference determination module acquires the test working condition record under normal electrical fire monitoring status, generates the current working condition identifier Kt based on the test working condition record, and determines the working condition reference vector Bt and the working condition boundary vector Et in the normal operation record. Step 2: The parameter deviation generation module uses the working condition reference vector Bt and the working condition boundary vector Et as a basis, compares them with the current parameter vector Xt, and generates the parameter deviation vector Qst. Step 3: The harmonic electrical correlation module correlates the parameters inside the parameter deviation vector Qst to generate the correlation factor Mdt; Step 4: The risk score generation module generates a basic score item FS based on the parameter deviation vector Qst, and uses the correlation factor Mdt to perform correlation correction on the basic score item to generate the risk score value RS and the risk status identifier YS. Step 5: The test feedback revision module extracts the test result identifier Zs from the test condition record and provides feedback based on the correspondence between the risk status identifier YS and the test result identifier Zs.

[0054] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A test system for monitoring harmonic electrical fires, characterized in that: It includes a working condition baseline determination module, a parameter deviation generation module, a harmonic electrical correlation module, a risk score generation module, and a test feedback revision module; The operating condition reference determination module acquires the test operating condition records under normal electrical fire monitoring conditions, generates the current operating condition identifier Kt based on the test operating condition records, and determines the operating condition reference vector Bt and the operating condition boundary vector Et in the normal operation records; The parameter deviation generation module uses the working condition reference vector Bt and the working condition boundary vector Et as a basis, compares them with the current parameter vector Xt, and generates the parameter deviation vector Qst. The harmonic electrical correlation module correlates the parameters that deviate from the vector Qst and generates a correlation factor Mdt. The risk score generation module generates a basic score item FS based on the parameter deviation vector Qst, and uses the correlation factor Mdt to perform correlation correction on the basic score item to generate the risk score value RS and the risk status identifier YS. The test feedback revision module extracts the test result identifier Zs from the test condition record and provides feedback based on the correspondence between the risk status identifier YS and the test result identifier Zs.

2. The test system for monitoring harmonic electrical fires according to claim 1, characterized in that: The operating condition baseline determination module includes an operating condition identifier generation unit and an operating condition baseline determination unit; The operating condition identifier generation unit obtains the test operating condition record corresponding to the t-th analysis window in the test system under normal electrical fire monitoring conditions, and extracts the load category identifier and operation phase identifier from the test operating condition record, and generates the current operating condition identifier Kt according to the combination relationship between the load category identifier and the operation phase identifier; The current operating condition identifier Kt is obtained by using the load category identifier as the first field, the running stage identifier as the second field, and generating the current operating condition identifier Kt in the form of a binary tuple. The same working condition benchmark determination unit uses the current working condition identifier Kt as the basis, and selects records that are the same as the current working condition identifier Kt from the existing normal operation records to form the same working condition normal operation record set Ωt; The formula for obtaining the normal operating record set Ωt under the same working conditions is: Ωt={r|Kr=Kt,Zr=N}; In the formula, Ωt specifically represents the set of normal operation records belonging to the same working condition as the t-th analysis window, r represents any historical analysis window record in the normal operation records, Kr represents the working condition identifier corresponding to the historical analysis window record r, Kt specifically represents the current working condition identifier corresponding to the t-th analysis window, Zr represents the test result identifier corresponding to the historical analysis window record r, and N represents the normal operation category; Among them, any historical analysis window record r in the normal operation record includes the following: total harmonic distortion rate Dr in historical analysis window record r, odd and even harmonic bias Pr in historical analysis window record r, harmonic phase offset Ar in historical analysis window record r, residual current fluctuation Lr in historical analysis window record r, load current change rate Cr in historical analysis window record r, and short-time impulse count Sr in historical analysis window record r; Based on the set of normal operating records Ωt under the same working conditions, the working condition reference vector Bt = (bD, bP, bA, bL, bC, bS) is calculated as follows: ; In the formula, bi represents the baseline value of parameter category i under the current operating condition identifier Kt, i represents the parameter category, and the value range is: total harmonic distortion rate D, odd and even harmonic offset P, harmonic phase offset A, residual current fluctuation L, load current change rate C, short-time impact count S, nt represents the number of records in the normal operation record set Ωt under the same operating condition, and x(r, i) represents the parameter value corresponding to parameter category i in the historical analysis window record r; Based on the normal operating record set Ωt and the operating condition reference vector Bt, the operating condition boundary vector Et = (eD, eP, eA, eL, eC, eS) is calculated as follows: ； In the formula, ei represents the allowable offset scale of parameter category i under the current working condition identifier Kt, εi represents the calculation resolution correction amount corresponding to parameter category i, γi represents the boundary conversion coefficient corresponding to parameter category i, and max represents taking the larger value between the calculation result and the calculation resolution correction amount.

3. The test system for monitoring harmonic electrical fires according to claim 2, characterized in that: The parameter deviation generation module includes a parameter alignment unit and a deviation generation unit; The parameter alignment unit receives the operating condition reference vector Bt and the operating condition boundary vector Et, and reads the data in the parameter vector Xt=(Dnt, Pnt, Ant, Lnt, Cnt, Snt) at the current time. It then establishes a one-to-one correspondence according to the parameter category and obtains the parameter alignment set HCt={(Dnt, bD, eD), (Pnt, bP, eP), (Ant, bA, eA), (Lnt, bL, eL), (Cnt, bC, eC), (Snt, bS, eS)}. In the formula, Dnt represents the total harmonic distortion rate at the current time, Pnt represents the odd and even harmonic bias at the current time, Ant represents the harmonic phase offset at the current time, Lnt represents the residual current fluctuation at the current time, Cnt represents the load current change rate at the current time, and Snt represents the short-time impulse count at the current time.

4. The test system for monitoring harmonic electrical fires according to claim 3, characterized in that: The deviation generation unit calculates the deviation of the parameters relative to the corresponding working condition reference value based on the obtained parameter alignment set HCt, and generates the parameter deviation vector Qst=(qD, qP, qA, qL, qC, qS). For parameter category i, the deviation is calculated as follows: ； In the formula, qi represents the parameter deviation of parameter category i, and xi represents the current parameter value corresponding to parameter category i in the parameter vector Xt at the current time; Wherein, qD represents the total harmonic distortion deviation, qP represents the odd and even harmonic bias deviation, qA represents the harmonic phase offset deviation, qL represents the residual current fluctuation deviation, qC represents the load current change rate deviation, and qS represents the short-time impulse count deviation. The deviation qi of parameter category i is analyzed to determine the degree of deviation, as follows: When qi=0, it means that the current parameter value of parameter category i is consistent with the corresponding working condition reference value; When 0 < qi < 1, it means that the current parameter value of parameter category i has an offset relative to the corresponding working condition reference value that does not reach the allowable offset scale. When qi=1, it means that the current parameter value of parameter category i is offset from the corresponding working condition reference value to the allowable offset scale.

5. The test system for monitoring harmonic electrical fires according to claim 4, characterized in that: The harmonic electrical correlation module includes a deviation matching unit and a correlation factor generation unit; The deviation pairing unit forms a first associated pairing with the total harmonic distortion rate deviation qD and the residual current fluctuation deviation qL in the parameter deviation vector Qst, forms a second associated pairing with the odd and even harmonic bias deviation qP and the load current change rate deviation qC, and forms a third associated pairing with the harmonic phase offset deviation qA and the short-time impulse count deviation qS. The first, second, and third associated pairings form an associated pairing set Gdt={(qD, qL), (qP, qC), (qA, qS)}.

6. The test system for monitoring harmonic electrical fires according to claim 5, characterized in that: The association factor generation unit is based on the association pairing set Gdt, and calculates three association sub-items respectively: the first association sub-item m1, the second association sub-item m2 and the third association sub-item m3, with the following formulas: m1=qDqL, m2=qPqC, m3=qAqS; In the formula, m1 specifically represents the degree of common deviation between the total harmonic distortion rate deviation and the residual current fluctuation deviation, m2 represents the degree of common deviation between the odd and even harmonic bias deviation and the load current change rate deviation, and m3 represents the degree of common deviation between the harmonic phase offset deviation and the short-time impulse count deviation, and 0≤m1≤1, 0≤m2≤1, 0≤m3≤1. Based on the three related sub-items, the correlation factor Mdt is calculated using the following formula: Mdt = a1m1 + a2m2 + a3m3; In the formula, a1 represents the first correlation coefficient, corresponding to the correlation pair (qD, qL); a2 represents the second correlation coefficient, corresponding to the correlation pair (qP, qC); a3 represents the third correlation coefficient, corresponding to the correlation pair (qA, qS), and 0≤a1≤1, 0≤a2≤1, 0≤a3≤1, a1+a2+a3=1.

7. A test system for monitoring harmonic electrical fires according to claim 6, characterized in that: The risk scoring generation module includes a weighted scoring generation unit and a scoring boundary generation and state mapping unit; The weighted scoring generation unit reads the set of marked test records for the same working condition corresponding to the current working condition identifier Kt; The set of labeled test records under the same working conditions is Φt={r|Or=Ot,Jr=Jt,Zr∈{N,H}}; In the formula, Or represents the load category identifier of the historical analysis window record r, Ot represents the load category identifier of the t-th analysis window, Jr represents the running stage identifier of the historical analysis window record r, Jt represents the running stage identifier of the t-th analysis window, N represents the normal operation category, and H represents the fault scenario category. The basic weight vector W = (wD, wP, wA, wL, wC, wS) is calculated based on the parameter separation degree between the normal operation category and the failure scenario category in the labeled test record set Φt under the same working conditions. The steps are as follows: S1. Let the set of parameter categories be I = {D, P, A, L, C, S}; S2. For any parameter category i∈I, first calculate the parameter deviation from the mean μ(i,N) under the normal operation category, and then calculate the parameter deviation from the mean μ(i,H) under the fault scenario category. S3. Based on the parameter deviation mean μ(i, N) under the normal operation category and the parameter deviation mean μ(i, H) under the fault scenario category, calculate and obtain the deviation discrete d(i, N) under the normal operation category and the deviation discrete d(i, H) under the fault scenario category. S4. First, calculate the class separation degree ηi of parameter class i, then normalize the class separation degree of each parameter class to obtain the basic weight wi, and obtain the basic weight vector W based on the basic weight wi. Based on the basic weight vector W, and combined with the current parameter deviation vector Qst, a basic scoring item FS is generated; The basic scoring item FS is obtained using the following formula: ; The risk score value RS is calculated by combining the basic scoring item FS with the correlation factor Mdt, as shown in the following formula; ； In the formula, min means that when the result after correlation correction exceeds 1, it is counted as 1; 100 means that the score result is converted into a score range of 0 to 100; and λ means the correlation correction coefficient.

8. The test system for monitoring harmonic electrical fires according to claim 7, characterized in that: The scoring boundary generation and state mapping unit compares the risk score value RS with the set first scoring boundary θ1 and second scoring boundary θ2 to obtain the risk state identifier YS. Risk status identifier YS is obtained through matching in the following way: When 0 ≤ RS < θ1, YS = Y1; When θ1≤RS<θ2, YS=Y2; When θ2≤RS≤100, YS=Y3; In the formula, Y1 represents the first risk state, Y2 represents the second risk state, and Y3 represents the third risk state.

9. A test system for monitoring harmonic electrical fires according to claim 8, characterized in that: The test feedback revision module includes a feedback relationship discrimination unit and a parameter revision feedback unit; The feedback relationship discrimination unit extracts the test result identifier Zs from the test condition record corresponding to the t-th analysis window, and maps it to the risk status identifier YS to generate the baseline revision mark cB and the score revision direction gs; Where Zs∈{N,H}; The benchmark revision flag cB is used to control whether the current parameter vector can enter the revision process of the operating condition benchmark vector Bt and the operating condition boundary vector E; ； In the formula, Zs=N indicates that the test result of the t-th analysis window is identified as the normal operation category, and Zs=H indicates that the test result of the t-th analysis window is identified as the fault scenario category; The score revision direction (gs) is used to control the revision direction of score weights and correlation factor correction coefficients; ； In the formula, gs=1 indicates that the fault scenario category is not mapped to the third risk state, gs=-1 indicates that the normal operation category is mapped to the second or third risk state, and gs=0 indicates that the correspondence between the risk state identifier YS and the test result identifier Zs is consistent. The parameter revision feedback unit revises the working condition baseline vector Bt, working condition boundary vector Et, scoring weight wi, and correlation factor correction coefficient λ according to the baseline revision mark cB and the scoring revision direction gs, and then feeds it back. The revision method for the operating condition reference vector Bt is as follows: ; In the formula, nbi represents the revised working condition baseline value of parameter category i, and ρB represents the working condition baseline revision coefficient; The method for revising the working condition boundary vector Et is as follows: ; In the formula, nei represents the revised boundary value of the working condition for parameter category i, and ρE represents the boundary revision coefficient for the working condition. The revision method for the scoring weight wi is as follows: ; In the formula, nwi represents the intermediate weight of parameter category i, ρW represents the weight revision coefficient, with a value range of 0 < ρW ≤ 0.1, and j represents any parameter category in the parameter category set; The intermediate weights of parameter category i are normalized to obtain the revised score weight znwi, as shown in the following formula: ； The revision method for the correlation factor correction coefficient λ is as follows: ; In the formula, ρλ represents the revision coefficient of the correlation factor correction coefficient, and its value ranges from 0 to 0.

1.

10. A test method for monitoring harmonic electrical fires, applied to the test system for monitoring harmonic electrical fires as described in any one of claims 1 to 9, characterized in that: Includes the following steps: Step 1: The working condition reference determination module acquires the test working condition record under normal electrical fire monitoring status, generates the current working condition identifier Kt based on the test working condition record, and determines the working condition reference vector Bt and the working condition boundary vector Et in the normal operation record. Step 2: The parameter deviation generation module uses the working condition reference vector Bt and the working condition boundary vector Et as a basis, compares them with the current parameter vector Xt, and generates the parameter deviation vector Qst. Step 3: The harmonic electrical correlation module correlates the parameters inside the parameter deviation vector Qst to generate the correlation factor Mdt; Step 4: The risk score generation module generates a basic score item FS based on the parameter deviation vector Qst, and uses the correlation factor Mdt to perform correlation correction on the basic score item to generate the risk score value RS and the risk status identifier YS. Step 5: The test feedback revision module extracts the test result identifier Zs from the test condition record and provides feedback based on the correspondence between the risk status identifier YS and the test result identifier Zs.

Images