Multi-source fusion safety early warning method and system for transformer substation inspection personnel

Abstract

The invention discloses a multi-source fusion safety early warning method and system for substation inspection personnel, and belongs to the technical field of power system operation and maintenance safety monitoring. The method comprises the following steps: determining an area where an inspector is located when the inspector inspects a transformer substation, and in the area, establishing an association relationship between motion human body characteristics of the inspector near an electric field of transformer substation equipment and changes of an induced electric field; performing routing inspection Markov chain modeling based on the association relationship, and estimating the routing inspection Markov chain and a transfer matrix to obtain an estimation result; and performing state risk assessment based on the estimation result, performing grade division on the assessed state risk according to a preset early warning grade judgment criterion, and sending out a corresponding early warning. According to the invention, the state risk of the inspection personnel in the inspection process can be effectively evaluated through the Markov chain, so that the safety of the inspection personnel in the operation process is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of power system operation and maintenance safety monitoring technology, and more specifically, to a multi-source fusion safety early warning method and system for substation inspection personnel. Background Technology

[0002] Currently, power system inspectors often wear portable electric field sensors in substations to monitor the surrounding electric field strength and trigger alarms when it exceeds a limit. Existing technology generally uses a single, fixed electric field threshold as the alarm criterion; that is, a danger warning is only issued when the sensor's detected value exceeds the set threshold. However, this method has the following significant drawbacks:

[0003] (1) Alarm delay, unable to handle momentary dangerous actions:

[0004] While the electric field may not exceed the threshold when the inspector is standing in certain positions, the electric field strength can increase instantaneously if the worker suddenly raises their hand, approaches the equipment, or changes their posture. Since raising or reaching out usually takes less than 0.3 seconds, traditional methods based on fixed thresholds often fail to detect this in time. By the time the electric field exceeds the limit and an alarm is triggered, it is too late to prevent danger.

[0005] (2) The location of staff and inspection routes were not taken into account:

[0006] Inspection personnel move through the substation in sequence according to the equipment. The electric field environment, operation actions and operating logic of different equipment have obvious temporal sequence. However, traditional technology only makes judgments based on the instantaneous values ​​of sensors and cannot combine the spatial location of the staff (such as transformers, disconnect switches and other areas) to make risk inferences.

[0007] (3) Unable to recognize action status:

[0008] Electric field sensors are essentially passively acquired single-value signals. Existing solutions do not have the ability to identify actions through changes in electric field characteristics, and cannot distinguish between situations with completely different risk levels, such as "walking close to the device," "running close to the device," and "standing while operating the device."

[0009] (4) Lacks the ability to predict future dangers:

[0010] The behavior of inspection personnel in substations exhibits strong regularity and a typical state transition sequence, such as "approaching equipment → standing up → raising hand to inspect → leaving." Existing technologies do not utilize the historical behavior of inspection personnel to establish a time-series model, and lack the ability to predict in advance "what might happen next in a dangerous state."

[0011] In summary, existing technologies cannot enable substation inspection personnel to conduct comprehensive risk assessments based on location-action-behavior sequences, nor can they provide early warnings of sudden dangerous postures, posing significant safety hazards. Summary of the Invention

[0012] To overcome the shortcomings of existing technologies, such as alarm lag, inability to identify actions, inability to integrate with inspection paths, and inability to predict dangerous behaviors, a multi-source fusion safety early warning method for substation inspection personnel is proposed, including:

[0013] Determine the area where the inspection personnel are inspecting the substation, and within that area, establish the correlation between the movement characteristics of the inspection personnel near the electric field of the substation equipment and the changes in the induced electric field;

[0014] Based on the aforementioned relationship, a Markov chain model for inspection is performed, and the Markov chain and transition matrix are estimated to obtain the estimation results.

[0015] Based on the estimation results, the state risk is assessed, and the assessed state risk is classified into levels according to the preset warning level judgment criteria, and corresponding warnings are issued.

[0016] Optionally, the area where the inspection personnel are located during the substation inspection can be determined, including:

[0017] After the camera's field of view fully covers the typical passages and the front area of ​​the equipment in the substation, the visual algorithm is used to determine the area where the inspection personnel are during their inspection, thus identifying the area where the inspection personnel are located during the substation inspection.

[0018] Optionally, establish the correlation between the moving human characteristics of inspection personnel near the electric field of substation equipment and changes in the induced electric field, including:

[0019] For typical near-electric scenarios in substations, a three-dimensional scene model is established. Through finite element simulation of the three-dimensional scene model, the electric field response characteristics of the electric field sensing sensors worn on the head and arms of the inspection personnel under different typical working actions and positions are obtained.

[0020] Based on the response characteristics and the speed and acceleration variation patterns of the inspection personnel during their movement, a judgment model for different movement states is established to determine the status of the inspection personnel.

[0021] Optional, typical near-electrical scenarios for substations include:

[0022] Transformer area scene, disconnector area scene, instrument transformer area scene, circuit breaker area scene and power distribution room area scene.

[0023] Optional decision models include:

[0024] G t = |dE / dt| < G1, indicating stationary / slow movement;

[0025] When G1 ≤ |dE / dt| < G2, it is determined as walking;

[0026] When |dE / dt| > G2, it is determined as running;

[0027] Where Gt is the electric field change rate, dt is the sampling period, E is the electric field modulus, and G1, G2 are determination thresholds.

[0028] Optionally, the patrol Markov chain modeling includes:

[0029] Denote the substation working space as The spatial positions of the staff at discrete times t = 0, 1, …, n are:

[0030] P t =(x t , y t , z t ) ∈ Ω

[0031] Then, the output of the wearable electric field sensor is the instantaneous electric field strength:

[0032] E t = E(P t , θ t )

[0033] Where θ t represents the human body posture;

[0034] ]>Denote the set of substation equipment as:

[0035] D = {1, 2, i…j, …}

[0036] Where the substation equipment includes: transformers, disconnectors, instrument transformers, circuit breakers, and switchgear rooms;

[0037] Therefore, the behavior of the patrol personnel is discretized into a finite state set:

[0038] X = S dev ∪ S path

[0039] Where the equipment state set: S dev = {S i,idle , S i,hand | i ∈ D}, S i,idle is staying / standing near equipment i, and S i,hand is raising a hand to operate near equipment i;

[0040] Where the path state set: S path = {P i→jwalk , P i→jrun | i, j ∈ D, i = j, (i → j) has a patrol path}, P i→jwalkTo walk from device i to device j, P i→jrun To run from device i to device j.

[0041] Optionally, Markov chains only allow the following typical transitions:

[0042] S i,idle Convert to S i,hand S i,idle Convert to P i→jwalk or P i→jrun ;

[0043] On the path, only the following typical transfers are allowed:

[0044] P i→jwalk Convert to S j,idle P i→jrun Convert to S j,idle .

[0045] Optionally, the inspection Markov chain and transition matrix are estimated, including:

[0046] If the staff state sequence {X} t If a chain follows a first-order discrete-time Markov chain, then:

[0047] P r {X t+1 =x j |X t =x i ,X t-1 ,…}=P r {X t+1 =x j |X t =x i}=P ij ,

[0048] in: Let P be the transition probability matrix. ij For P r The probability of transitioning from state i to state j;

[0049] Historical inspection routes Number of transfers:

[0050] N ij =#{t∣X t =x i ,X t+1 =x j},

[0051] Then: the transition probability estimated by maximum likelihood is:

[0052]

[0053] Among them, N ij is the number of transfers from i to j in the statistical history.

[0054] Optionally, for the calculation formula of the transition probability of maximum likelihood estimation, exponential weighted update is adopted:

[0055]

[0056] Among them, is the transition probability obtained by statistics within a recent time window, λ controls the weight of new data, is the value of the element P in the transition matrix in the next inspection after weighted update, ij and is the transition probability obtained by statistics within a recent time window.

[0057] Optionally, the calculation formula for the assessment of state risk is as follows:

[0058]

[0059] Among them, R total (t) is the comprehensive risk index, α, β, γ are coefficients, and the sum of α, β, γ is 1. E lit and G lit are the standard limits respectively, and E t and G t are the measured real-time electric field value and the real-time measured electric field gradient value respectively. R pred (t) is the predicted risk score.

[0060] Optionally, the decision formula for grading the assessed state risk is as follows:

[0061]

[0062] Among them, 0 < p1 < p2, 0 < λ1 < λ2 are threshold parameters;

[0063] Among them, 0 represents safety, 1 represents potential danger, and 2 represents danger.

[0064] On the other hand, the present invention also proposes a multi-source fusion safety warning system for substation inspection personnel, including:

[0065] An association unit, configured to determine the area where the inspection personnel are located during substation inspection, and establish an association relationship between the moving human body characteristics of the inspection personnel near the electric field of substation equipment and the change of the induced electric field in the area;

[0066] An estimation unit, configured to perform patrol Markov chain modeling based on the association relationship, estimate the patrol Markov chain and the transition matrix, and obtain an estimation result;

[0067] An output unit, configured to evaluate the state risk based on the estimation result, classify the evaluated state risk according to a preset warning level determination criterion, and issue a corresponding warning.

[0068] Optionally, determining the area where the inspection personnel are located during the substation inspection includes:

[0069] After the typical channels and the front areas of the equipment in the substation are fully covered by the camera's field of view, the area where the inspection personnel are located during the inspection is judged through a vision algorithm to determine the area where the inspection personnel are located during the substation inspection.

[0070] Optionally, establishing the correlation between the moving human body characteristics of the inspection personnel near the electric field of the substation equipment and the change of the induced electric field includes:

[0071] For the typical near-electric scenarios of the substation, a three-dimensional scene model is established, and through the finite element simulation of the three-dimensional scene model, the electric field response characteristics of the electric field induction sensors worn on the heads and arms of the inspection personnel at different typical operation actions and different positions are obtained;

[0072] According to the response characteristics and the variation laws of the speed and acceleration of the inspection personnel in the moving state, a determination model for different moving states is established to determine the state of the inspection personnel.

[0073] Optionally, the typical near-electric scenarios of the substation include:

[0074] Transformer area scene, disconnector area scene, instrument transformer area scene, circuit breaker area scene, and distribution room area scene.

[0075] Optionally, the determination model includes:

[0076] G t =|dE / dt| < G1, it is determined as stationary / slow moving;

[0077] G1 ≤ |dE / dt| < G2, it is determined as walking;

[0078] |dE / dt| > G2, it is determined as running;

[0079] Where, Gt is the electric field change rate, dt is the sampling period, E is the electric field modulus, and G1, G2 are determination thresholds.

[0080] Optionally, the inspection Markov chain modeling includes:

[0081] Denote the substation working space as The spatial positions of the staff at discrete times t = 0, 1,..., n are:

[0082] P t =(x t ,y t ,z t )∈Ω

[0083] Therefore, the wearable electric field sensor outputs the instantaneous electric field strength:

[0084] E t =E(P t ,θ t )

[0085] Where, θ t Indicates human posture;

[0086] The collection of substation equipment is denoted as:

[0087] D = {1, 2, i, ..., j, ...}

[0088] The substation equipment includes: transformers, disconnect switches, instrument transformers, circuit breakers, and power distribution rooms;

[0089] Therefore, the behavior of the inspection personnel can be discretized into a finite set of states:

[0090] X = S dev ∪S path

[0091] Among them, the device state set is: S dev ={S i,idle ,S i,hand |i∈D},S i,idle To stay / stand near device i, S i,hand To operate by raising your hand near device i;

[0092] Among them, the path state set: S path ={P i→jwalk ,P i→jrun |i,j∈D,i=j,(i→j)have an inspection path},P i→jwalk To walk from device i to device j, P i→jrun To run from device i to device j.

[0093] Optionally, Markov chains only allow the following typical transitions:

[0094] S i,idle Convert to S i,hand S i,idle Convert to P i→jwalk or P i→jrun ;

[0095] On the path, only the following typical transfers are allowed:

[0096] P i→jwalk Convert to Sj,idle P i→jrun Convert to S j,idle .

[0097] Optionally, the inspection Markov chain and transition matrix are estimated, including:

[0098] If the staff state sequence {X} t If a chain follows a first-order discrete-time Markov chain, then:

[0099] P r {X t+1 =x j |X t =x i ,X t-1 ,…}=P r {X t+1 =x j |X t =x i}=P ij ,

[0100] in: Let P be the transition probability matrix. ij For P r The probability of transitioning from state i to state j;

[0101] Historical inspection routes Number of transfers:

[0102] N ij =#{t∣X t =x i ,X t+1 =x j},

[0103] Then: the transition probability estimated by maximum likelihood is:

[0104]

[0105] Where, N ij This is to count the number of transitions from i to j in history.

[0106] Optionally, the formula for calculating the transition probability of the maximum likelihood estimate is updated using an exponentially weighted approach:

[0107]

[0108] in, The transition probabilities are statistically obtained within a recent window, and λ controls the weight of new data. For the element P in the transition matrix of the next inspection after weighted update ij The value, It is the transition probability statistically obtained within the most recent time window.

[0109] Optionally, the calculation formula for the evaluation of state risk is as follows:

[0110]

[0111] Among them, R total (t) is the comprehensive risk index, α, β, and γ are coefficients, and the sum of α, β, and γ is 1. E lit and G lit are the standard limits respectively. E t and G t are the measured real-time electric field value and the real-time measured electric field gradient value respectively. R pred (t) is the predicted risk score.

[0112] Optionally, the determination formula for grading the evaluated state risk is as follows:

[0113]

[0114] Among them, 0 < p1 < p2, 0 < λ1 < λ2 are threshold parameters;

[0115] Among them, 0 represents safety, 1 represents potential danger, and 2 represents danger.

[0116] On the other hand, the present invention also provides a computing device, including: one or more processors;

[0117] The processor is used to execute one or more programs;

[0118] When the one or more programs are executed by the one or more processors, the method as described above is implemented.

[0119] On the other hand, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed, the method as described above is implemented.

[0120] Compared with the prior art, the beneficial effects of the present invention are:

[0121] This invention provides a multi-source fusion safety early warning method for substation inspection personnel, comprising: determining the area where the inspection personnel are located during substation inspections; establishing a correlation between the movement characteristics of the inspection personnel and changes in the induced electric field near the substation equipment's electric field within the area; based on the correlation, modeling an inspection Markov chain, estimating the inspection Markov chain and transition matrix, and obtaining the estimation result; based on the estimation result, assessing the state risk, classifying the assessed state risk according to a preset early warning level judgment criterion, and issuing corresponding early warnings. This invention can effectively assess the state risk of inspection personnel during the inspection process through Markov chains, thereby ensuring the safety of inspection personnel during operations. Attached Figure Description

[0122] Figure 1 This is a flowchart of the method of the present invention;

[0123] Figure 2 This is a schematic diagram of a Markov chain in an embodiment of the method of the present invention;

[0124] Figure 3 This is a structural diagram of the system of the present invention. Detailed Implementation

[0125] Exemplary embodiments of the invention will now be described with reference to the accompanying drawings. However, the invention may be embodied in many different forms and is not limited to the embodiments described herein. These embodiments are provided to fully and completely disclose the invention and to fully convey its scope to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the drawings is not intended to limit the invention. In the drawings, the same units / elements are referred to by the same reference numerals.

[0126] Unless otherwise stated, the terms used herein (including technical terms) have their common meaning as understood by one of ordinary skill in the art. Furthermore, it is understood that terms defined in commonly used dictionaries should be understood to have a meaning consistent with the context of their relevant field, and not to be interpreted as having an idealized or overly formal meaning.

[0127] Example 1:

[0128] This invention proposes a multi-source fusion safety early warning method S100 for substation inspection personnel, such as... Figure 1 As shown, it includes:

[0129] S101. Determine the area where the inspection personnel are inspecting the substation, and establish the correlation between the movement characteristics of the inspection personnel and the changes in the induced electric field near the electric field of the substation equipment in the area.

[0130] S102. Based on the association relationship, perform patrol Markov chain modeling, estimate the patrol Markov chain and the transition matrix, and obtain the estimation result;

[0131] S103. Based on the estimation result, conduct state risk assessment, classify the evaluated state risk according to the preset warning level judgment criterion, and issue corresponding warnings.

[0132] Among them, determining the area where the patrol personnel are located during substation patrol includes:

[0133] After the typical channels and the front areas of the equipment in the substation are fully covered by the camera's field of view, use a vision algorithm to judge the area where the patrol personnel are located during patrol, so as to determine the area where the patrol personnel are located during substation patrol.

[0134] Among them, establishing the association relationship between the moving human body characteristics of the patrol personnel near the electric field of the substation equipment and the change of the induced electric field includes:

[0135] For the typical near-electric scenarios of the substation, establish a three-dimensional scene model, and through the finite element simulation of the three-dimensional scene model, obtain the electric field response characteristics of the electric field induction sensors worn on the heads and arms of the patrol personnel under different typical operation actions and different positions;

[0136] According to the response characteristics and the speed and acceleration change rules of the patrol personnel in the moving state, establish a judgment model for different moving states to judge the state of the patrol personnel.

[0137] Among them, the typical near-electric scenarios of the substation include:

[0138] Transformer area scene, disconnector area scene, instrument transformer area scene, circuit breaker area scene and distribution room area scene.

[0139] Among them, the judgment model includes:

[0140] G t =|dE / dt| < G1, judged as stationary / slow moving;

[0141] G1 ≤ |dE / dt| < G2, judged as walking;

[0142] |dE / dt| > G2, judged as running;

[0143] Among them, Gt is the electric field change rate, dt is the sampling period, E is the electric field modulus, and G1 and G2 are judgment thresholds.

[0144] Among them, the patrol Markov chain modeling includes:

[0145] Denote the substation working space as The spatial locations of the staff at discrete time t = 0, 1, ..., n are:

[0146] P t =(x t ,y t ,z t )∈Ω

[0147] Therefore, the wearable electric field sensor outputs the instantaneous electric field strength:

[0148] E t =E(P t ,θ t )

[0149] Where, θ t Indicates human posture;

[0150] The collection of substation equipment is denoted as:

[0151] D = {1, 2, i, ..., j, ...}

[0152] The substation equipment includes: transformers, disconnect switches, instrument transformers, circuit breakers, and power distribution rooms;

[0153] Therefore, the behavior of the inspection personnel can be discretized into a finite set of states:

[0154] X = S dev ∪S path

[0155] Among them, the device state set is: S dev ={S i,idle ,S i,hand |i∈D},S i,idle To stay / stand near device i, S i,hand To operate by raising your hand near device i;

[0156] Among them, the path state set: S path ={P i→jwalk ,P i→jrun |i,j∈D,i=j,(i→j)have an inspection path},P i→jwalk To walk from device i to device j, P i→jrun To run from device i to device j.

[0157] Markov chains only allow the following typical transitions:

[0158] S i,idle Convert to S i,hand S i,idle Convert to P i→jwalk or P i→jrun ;

[0159] On the path, only the following typical transfers are allowed:

[0160] P i→jwalk Convert to S j,idle P i→jrun Convert to S j,idle .

[0161] The estimation of the Markov chain and transition matrix during inspection includes:

[0162] If the staff state sequence {X} t If a chain follows a first-order discrete-time Markov chain, then:

[0163] P r {X t+1 =x j |X t =x i ,X t-1 ,…}=P r {X t+1 =x j |X t =x i}=P ij ,

[0164] in: Let P be the transition probability matrix. ij For P r The probability of transitioning from state i to state j;

[0165] Historical inspection routes Number of transfers:

[0166] N ij =#{t∣X t =x i ,X t+1 =x j},

[0167] Then: the transition probability estimated by maximum likelihood is:

[0168]

[0169] Where, N ij This is to count the number of transitions from i to j in history.

[0170] The formula for calculating the transition probability of the maximum likelihood estimate uses an exponentially weighted update:

[0171]

[0172] in, The transition probabilities are statistically obtained within a recent window, and λ controls the weight of new data. is the value of the element P in the transition matrix during the next inspection after weighted update ij . is the transition probability obtained by statistics within the recent time window.

[0173] Among them, the calculation formula for the assessment of state risk is as follows:

[0174]

[0175] Among them, R total (t) is the comprehensive risk index, α, β, γ are coefficients, and the sum of α, β, γ is 1. E lit and G lit are the standard limits respectively. E t and G t are the measured real-time electric field value and the real-time measured electric field gradient value respectively. R pred (t) is the predicted risk score.

[0176] Among them, the judgment formula for grading the evaluated state risk is as follows:

[0177]

[0178] Among them, 0 < p1 < p2, 0 < λ1 < λ2 are threshold parameters;

[0179] Among them, 0 represents safety, 1 represents potential danger, and 2 represents danger.

[0180] The following further illustrates the present invention in combination with specific cases:

[0181] Step 1: Obtain the area where the inspection personnel are located in the substation area:

[0182] Divide the substation into equipment areas and passage areas, install fixed cameras at the main transformer, switchyard, busbar area, passage corners, etc. The camera's field of view covers typical passages and the front areas of equipment, and judge the area where the personnel are located based on the existing monitoring devices in the substation and using vision algorithms.

[0183] Step 2: Establish the correlation between the characteristics of moving human bodies near the electric field of substation equipment and the change of the induced electric field:

[0184] For typical near-electricity scenarios in a substation (including areas such as transformers, disconnectors, instrument transformers, circuit breakers, and distribution rooms), first establish the corresponding three-dimensional scene model. Considering typical operation actions of inspection personnel approaching equipment along the passage on foot or running, raising the knife switch or checking the contact points beside the equipment, etc., based on finite element simulation, obtain the electric field response characteristics of the electric field induction sensors worn on the heads and arms of inspection personnel under different actions and different positions. At the same time, according to the speed and acceleration change rules of the human body in motion states such as walking and running, establish the determination of different moving states through the different frequencies and amplitudes of the electric field signal of the head sensor changing with time. The sampling period is specified as dt, and the electric field change rate is approximately:

[0185] G t =|dE / dt| < G1, it is determined as stationary / slow movement; G1 ≤ |dE / dt| < G2, it is determined as walking; |dE / dt| > G2, it is determined as running; when the known position is almost unchanged, if the electric field modulus value E obtained by the arm sensor undergoes a jump exceeding a certain proportion (for example, ΔE / E > α), it is determined as a raising hand operation near the equipment.

[0186] Among them, G1 and G2 are determination thresholds. Through the above model, the current state of the inspection personnel, including walking, running, and raising hand operations near the equipment, can be identified from the real-time signal changes of the electric field sensor.

[0187] Step 3: Modeling of the inspection Markov chain:

[0188] The working space of the substation is denoted as The spatial positions of the staff at discrete times t = 0, 1,... are:

[0189] P t =(x t , y t , z t ) ∈ Ω

[0190] The output of the wearable electric field sensor is the instantaneous electric field intensity:

[0191] E t =E(P t , θ t [[ID=۳۹]]

[0192] Among them, θ t represents the human body posture.

[0193] The set of substation equipment is denoted as:

[0194] D = {1, 2, i…j,…}

[0195] Respectively represent transformers, disconnectors, instrument transformers, circuit breakers, distribution rooms, etc.

[0196] To simultaneously characterize equipment position and operational actions, worker behavior is discretized into a finite set of states:

[0197] X = S dev ∪S path

[0198] Device state set: S dev ={S i,idle ,S i,hand |i∈D},S i,idle This indicates staying / standing near device i; S i,hand This indicates that you are raising your hand to operate near device i.

[0199] Path state set: S path ={P i→jwalk ,P i→jrun |i,j∈D,i=j,(i→j)have an inspection path},P i→jwalk Walk from device i to device j; P i→jrun Run from device i to device j.

[0200] To satisfy physical feasibility, only the following typical transfers are allowed: S i,idle Convert to S i,hand S i,idle Convert to P i→jwalk Or P i→jrun On the path, only the following typical transitions are allowed: P i→jwalk Convert to S j,idle ;P i→jrun Convert to S j,idle .

[0201] Specifically, such as Figure 2 The examples shown are devices 1–5, representing the inspection sequence:

[0202] The inspector walks from the transformer (1) to the disconnector (2) →

[0203] Raise your hand near the disconnect switch → S 2,hand ;

[0204] Running from the disconnector to the circuit breaker (4) →

[0205] Stay near the circuit breaker: →S 4,idle ;

[0206] Assuming from S 4,idle To S 4,hand If the historical probability is high (such as when raising your hand is required for operations like pulling a knife switch), the system will automatically calculate:

[0207]

[0208] If the probability is high, an early warning will be triggered, even if the electric field value is still within a safe range.

[0209] Step 4: Markov chain and transition matrix estimation:

[0210] Assume a sequence of worker states {X} t It follows a first-order discrete-time Markov chain:

[0211] P r {X t+1 =x j |X t =x i ,X t-1 ,…}=P r {X t+1 =x j |X t =x i}=P ij ,

[0212] in Let be the transition probability matrix.

[0213] Historical inspection routes Number of transfers:

[0214] N ij =#{t∣X t =x i ,X t+1 =x j}

[0215] The transition probability estimated by the maximum likelihood is:

[0216]

[0217] To enable the model to continuously adapt to inspection behavior, an exponentially weighted update method can be used:

[0218]

[0219] in The transition probability is obtained statistically within the window over a recent period, and λ controls the weight of new data.

[0220] Step 5: State Risk Assessment Model:

[0221] First, a real-time threshold and gradient risk assessment is performed, based on the limit E given by relevant standards. lit G lit Real-time measurement of E t G t The normalized quantity is defined later:

[0222]

[0223] Then, based on the Markov chain, predict the risk of behavior. Let the current state be X t = x i , and its one-hot vector representation is The state probability vectors for one-step prediction and two-step prediction are respectively:

[0224] π(1) = δ xi P

[0225] π(2) = δ xi P 2

[0226] Define the set of high-risk states χ H :

[0227]

[0228] Taking into account both one-step and two-step predictions, the predicted risk score can be defined as:

[0229]

[0230] The system issues a level 1-2 early warning in advance. This mechanism can identify dangerous actions such as "raising hands and running towards high-voltage equipment" in advance, providing an early warning several seconds earlier than a single threshold judgment.

[0231] Finally, construct the comprehensive risk index R total (t):

[0232]

[0233] where the sum of the parameters α, β, and γ is 1.

[0234] Step 6: Set the warning level determination:

[0235] According to R total (t) and the protective condition, trigger the warning level L(t) ∈ {0, 1, 2}:

[0236]

[0237] where 0 < p1 < p2, 0 < λ1 < λ2 are threshold parameters.

[0238] Trigger level definitions: Level 0 (Safe): Both the electric field and gradient are within a safe range, and the probability of entering a high-risk state in the short term is small; Level 1 (Potential Danger): The current electric field / gradient is close to the limit, or the Markov chain predicts that the worker is likely to evolve into a high-risk device / posture; Level 2 (Danger): The electric field is close to or exceeds the limit, or there is a very high probability that the worker will enter a high-risk posture in the next step (such as raising their hand to operate in front of high-voltage equipment), and an immediate voice + sound and light forced alarm is required.

[0239] Example 2:

[0240] Furthermore, this invention also proposes a multi-source fusion safety early warning system 200 for substation inspection personnel, such as... Figure 3 As shown, it includes:

[0241] The association unit 201 is used to determine the area where the inspection personnel are inspecting the substation, and to establish the correlation between the movement characteristics of the inspection personnel and the changes in the induced electric field near the electric field of the substation equipment in the area.

[0242] The estimation unit 202 is used to model the inspection Markov chain based on the association relationship, estimate the inspection Markov chain and the transition matrix, and obtain the estimation result.

[0243] The output unit 203 is used to assess the state risk based on the estimation result, classify the assessed state risk into levels according to the preset warning level judgment criteria, and issue corresponding warnings.

[0244] This includes determining the area where inspection personnel are located during substation inspections, including:

[0245] After the camera's field of view fully covers the typical passages and the front area of ​​the equipment in the substation, the visual algorithm is used to determine the area where the inspection personnel are during their inspection, thus identifying the area where the inspection personnel are located during the substation inspection.

[0246] This includes establishing the correlation between the movement characteristics of inspection personnel near the electric field of substation equipment and changes in the induced electric field, including:

[0247] For typical near-electric scenarios in substations, a three-dimensional scene model is established. Through finite element simulation of the three-dimensional scene model, the electric field response characteristics of the electric field sensing sensors worn on the head and arms of the inspection personnel under different typical working actions and positions are obtained.

[0248] Based on the response characteristics and the speed and acceleration variation patterns of the inspection personnel during their movement, a judgment model for different movement states is established to determine the status of the inspection personnel.

[0249] Typical near-electricity scenarios for substations include:

[0250] Transformer area scene, disconnector area scene, current transformer area scene, circuit breaker area scene, and distribution room area scene.

[0251] Among them, the determination model includes:

[0252] G t = |dE / dt| < G1, it is determined as stationary / slow moving;

[0253] G1 ≤ |dE / dt| < G2, it is determined as walking;

[0254] |dE / dt| > G2, it is determined as running;

[0255] Among them, Gt is the electric field change rate, dt is the sampling period, E is the electric field modulus, and G1 and G2 are determination thresholds.

[0256] Among them, the patrol Markov chain modeling includes:

[0257] The working space of the substation is denoted as The spatial positions of the staff at discrete times t = 0, 1,..., n are:

[0258] P t = (x t , y t , z t ) ∈ Ω

[0259] Then, the output of the wearable electric field sensor is the instantaneous electric field intensity:

[0260] E t = E(P t , θ t )

[0261] Among them, θ t represents the human body posture;

[0262] The set of substation equipment is denoted as:

[0263] D = {1, 2, i... j,...}

[0264] Among them, the substation equipment includes: transformers, disconnectors, current transformers, circuit breakers, and distribution rooms;

[0265] Therefore, the behavior of the patrol personnel is discretized into a finite state set:

[0266] X = S dev ∪ S path

[0267] Among them, the equipment state set: S dev = {S i,idle , S i,hand|i∈D},S i,idle To stay / stand near device i, S i,hand To operate by raising your hand near device i;

[0268] Among them, the path state set: S path ={P i→jwalk ,P i→jrun |i,j∈D,i=j,(i→j)have an inspection path},P i→jwalk To walk from device i to device j, P i→jrun To run from device i to device j.

[0269] Markov chains only allow the following typical transitions:

[0270] S i,idle Convert to S i,hand S i,idle Convert to P i→jwalk or P i→jrun ;

[0271] On the path, only the following typical transfers are allowed:

[0272] P i→jwalk Convert to S j,idle P i→jrun Convert to S j,idle .

[0273] The estimation of the Markov chain and transition matrix during inspection includes:

[0274] If the staff state sequence {X} t If a chain follows a first-order discrete-time Markov chain, then:

[0275] P r {X t+1 =x j |X t =x i ,X t-1 ,…}=P r {X t+1 =x j |X t =x i}=P ij ,

[0276] in: Let P be the transition probability matrix. ij For P r The probability of transitioning from state i to state j;

[0277] Historical inspection routes Number of transfers:

[0278] Nij = #{t | X t = x i , X t+1 = x j},

[0279] Then: The transition probability of the maximum likelihood estimate is as follows:

[0280]

[0281] where N ij is the number of transitions from i to j in the statistical history.

[0282] where, for the calculation formula of the transition probability of the maximum likelihood estimate, exponential weighted update is adopted:

[0283]

[0284] where is the transition probability obtained by statistics within the most recent time window, λ controls the weight of new data, is the value of the element P ij in the transition matrix in the next inspection after weighted update, is the transition probability obtained by statistics within the most recent time window.

[0285] where, the calculation formula for the assessment of state risk is as follows:

[0286]

[0287] where R total (t) is the comprehensive risk index, α, β, γ are coefficients, and the sum of α, β, γ is 1, E lit and G lit are the standard limits respectively, E t and G t are the measured real-time electric field value and the real-time measured electric field gradient value respectively, R pred (t) is the predicted risk score.

[0288] where, the decision formula for grading the evaluated state risk is as follows:

[0289]

[0290] where 0 < p1 < p2, 0 < λ1 < λ2 are threshold parameters;

[0291] where 0 represents safety, 1 represents potential danger, and 2 represents danger.

[0292] This invention enables early warning: by predicting the inspector's actions several steps ahead using Markov chains, it provides early warning before dangerous actions occur. It significantly reduces misjudgments and missed judgments caused by sudden hand gestures or sudden approach to equipment: through action recognition and electric field gradient judgment, it can identify dangerous behaviors such as rapid approach and running, achieving dynamic early warning. It unifies location-action-sequence information into a single state model, improving the accuracy of risk assessment and overcoming the shortcomings of traditional solutions that rely solely on electric field thresholds. It possesses self-learning capabilities: the model can automatically update state transition probabilities based on daily inspection trajectories, adapting to the habits of different inspectors and different substation layouts. It improves inspection safety and operational standardization, achieving "action-level" fine-grained safety monitoring, and can be widely applied to the management of substation maintenance personnel.

[0293] Example 3:

[0294] Based on the same inventive concept, this invention also provides a computer device, which includes a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to implement corresponding method flows or corresponding functions, thereby implementing the steps of the methods in the above embodiments.

[0295] Example 4:

[0296] Based on the same inventive concept, this invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of the method in the above embodiments.

[0297] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0298] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0299] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0300] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0301] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0302] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A multi-source fusion safety early warning method for substation inspection personnel, characterized in that, Including: Determine the area where the inspection personnel are located during the inspection of the substation. Under this area, establish the correlation between the moving human body characteristics of the inspection personnel near the electric field of the substation equipment and the change of the induced electric field; Based on the correlation, conduct patrol Markov chain modeling, and estimate the patrol Markov chain and the transition matrix to obtain the estimation result; Based on the estimation result, conduct state risk assessment, and according to the preset early warning level judgment criterion, classify the evaluated state risk and issue the corresponding early warning.

2. The multi-source fusion security early warning method according to claim 1, characterized in that, The determination of the area where the inspection personnel are located during the inspection of the substation includes: After the camera field of view fully covers the typical channels and the front area of the equipment in the substation, use the vision algorithm to judge the area where the inspection personnel are located during the inspection, so as to determine the area where the inspection personnel are located during the inspection of the substation.

3. The multi-source fusion security early warning method according to claim 1, characterized in that, The establishment of the correlation between the moving human body characteristics of the inspection personnel near the electric field of the substation equipment and the change of the induced electric field includes: For the typical near-electric scenarios of the substation, establish a three-dimensional scene model, and through the finite element simulation of the three-dimensional scene model, obtain the electric field response characteristics of the electric field induction sensors worn on the heads and arms of the inspection personnel under different typical operation actions and different positions; According to the response characteristics and the variation laws of the speed and acceleration of the inspection personnel in the moving state, establish a determination model for different moving states to determine the state of the inspection personnel.

4. The multi-source fusion security early warning method according to claim 3, characterized in that, The typical near-electric scenarios of the substation include: Transformer area scene, disconnector area scene, current transformer area scene, circuit breaker area scene and distribution room area scene.

5. The multi-source fusion security early warning method according to claim 3, characterized in that, The determination model includes: G t = |dE / dt| < G1, indicating stationary / slow movement; When G1≤∣dE / dt∣<G2, it is determined as walking; When ∣dE / dt∣>G2, it is determined as running; Where, Gt is the electric field change rate, dt is the sampling period, E is the electric field modulus, and G1 and G2 are determination thresholds.

6. The multi-source fusion security early warning method according to claim 1, characterized in that, The patrol Markov chain modeling includes: The substation workspace is denoted as The spatial locations of the staff at discrete time t = 0, 1, ..., n are: P t =(x t ,y t ,z t )∈Ω Then, the output of the wearable electric field sensor is the instantaneous electric field intensity: AND t =E(P t ,θ t ) Where, θ t Indicates human posture; Denote the set of substation equipment as: D={1,2,i…j,…} Where, the substation equipment includes: transformers, disconnectors, current transformers, circuit breakers and distribution rooms; Therefore, the behavior of the inspection personnel is discretized into a finite state set: X=S dev ∪S path Among them, the device state set is: S dev ={S i,idle ,S i,hand |i∈D},S i,idle To stay / stand near device i, S i,hand To operate by raising your hand near device i; Among them, the path state set: S path ={P i→jwalk ,P i→jrun |i,j∈D,i=j,(i→j)have an inspection path},P i→jwalk To walk from device i to device j, P i→jrun To run from device i to device j.

7. The multi-source fusion security early warning method according to claim 6, characterized in that, The Markov chain only allows the following typical transitions: S i,idle Convert to S i,hand S i,idle Convert to P i→jwalk or P i→jrun ; On the path, only the following typical transitions are allowed: P i→jwalk Convert to S j,idle P i→jrun Convert to S j,idle .

8. The multi-source fusion security early warning method according to claim 1, characterized in that, The estimation of the patrol Markov chain and the transition matrix includes: If the staff state sequence {X} t If a chain follows a first-order discrete-time Markov chain, then: P r {X t+1 =x j ∣X t =x i ,X t-1 ,…}=P r {X t+1 =x j ∣X t =x i }=P ij , in: Let P be the transition probability matrix. ij For P r The probability of transitioning from state i to state j; Historical inspection routes Number of transfers: N ij =#{t∣X t =x i ,X t+1 =x j }, Then: the transition probability of the maximum likelihood estimation is: Where, N ij This is to count the number of transitions from i to j in history.

9. The multi-source fusion security early warning method according to claim 8, characterized in that, For the calculation formula of the transition probability of the maximum likelihood estimation, use exponential weighted update: in, The transition probabilities are statistically obtained within a recent window, and λ controls the weight of new data. For the element P in the transition matrix of the next inspection after weighted update ij The value, The transition probability is the statistical result obtained within a recent window.

10. The multi-source fusion security early warning method according to claim 1, characterized in that, The calculation formula of the evaluation of the state risk is as follows: Among them, R total (t) is the comprehensive risk index, α, β, and γ are coefficients, and the sum of α, β, and γ is 1. lit and G lit These are the standard limits, E t and G t These represent the measured real-time electric field value and the real-time measured electric field gradient value, respectively, R. pred (t) represents the predicted risk score.

11. The multi-source fusion security early warning method according to claim 10, characterized in that, The judgment formula for classifying the evaluated state risk is as follows: Where, 0<p1<p2, 0<λ1<λ are threshold parameters; Where, 0 represents safety, 1 represents potential danger, and 2 represents danger.

12. A multi-source fusion safety early warning system for substation inspection personnel, characterized in that, Including: The correlation unit is used to determine the area where the inspection personnel are located during the inspection of the substation. Under this area, establish the correlation between the moving human body characteristics of the inspection personnel near the electric field of the substation equipment and the change of the induced electric field; The estimation unit is used to conduct patrol Markov chain modeling based on the correlation, and estimate the patrol Markov chain and the transition matrix to obtain the estimation result; An output unit for evaluating the state risk based on the estimation result, classifying the evaluated state risk according to a preset warning level determination criterion, and issuing a corresponding warning.

13. The multi-source fusion security early warning system according to claim 12, characterized in that, Determining the area where the inspection personnel are located during the inspection in the substation, including: After the typical channels and the front areas of the equipment in the substation are comprehensively covered by the camera's field of view, the area where the inspection personnel are located during the inspection is determined through a vision algorithm.

14. The multi-source fusion security early warning system according to claim 12, characterized in that, Establishing the correlation between the moving human body characteristics of the inspection personnel near the electric field of the substation equipment and the change of the induced electric field, including: For the typical near-electric scenarios of the substation, a three-dimensional scene model is established, and through the finite element simulation of the three-dimensional scene model, the electric field response characteristics of the electric field induction sensors worn on the heads and arms of the inspection personnel under different typical operation actions and different positions are obtained; According to the response characteristics and the speed and acceleration change rules of the inspection personnel in the moving state, a determination model for different moving states is established to determine the state of the inspection personnel.

15. The multi-source fusion security early warning system according to claim 14, characterized in that, The typical near-electric scenarios of the substation include: Transformer area scenario, disconnector area scenario, current transformer area scenario, circuit breaker area scenario, and distribution room area scenario.

16. The multi-source fusion security early warning system according to claim 14, characterized in that, The determination model includes: G t = |dE / dt| < G1, indicating stationary / slow movement; When G1≤∣dE / dt∣<G2, it is determined as walking; When ∣dE / dt∣>G2, it is determined as running; Where, Gt is the electric field change rate, dt is the sampling period, E is the electric field modulus, and G1 and G2 are determination thresholds.

17. The multi-source fusion security early warning system according to claim 12, characterized in that, Modeling the inspection Markov chain, including: The substation workspace is denoted as The spatial locations of the staff at discrete time t = 0, 1, ..., n are: P t =(x t ,y t ,z t )∈Ω Then, the output of the wearable electric field sensor is the instantaneous electric field intensity: AND t =E(P t ,θ t ) Where, θ t Indicates human posture; Denote the set of substation equipment as: D={1,2,i…j,…} Where, the substation equipment includes: transformers, disconnectors, current transformers, circuit breakers, and distribution rooms; Therefore, the behavior of the inspection personnel is discretized into a finite state set: X=S dev ∪S path Among them, the device state set is: S dev ={S i,idle ,S i,hand |i∈D},S i,idle To stay / stand near device i, S i,hand To operate by raising your hand near device i; Among them, the path state set: S path ={P i→jwalk ,P i→jrun |i,j∈D,i=j,(i→j)have an inspection path},P i→jwalk To walk from device i to device j, P i→jrun To run from device i to device j.

18. The multi-source fusion security early warning system according to claim 17, characterized in that, The Markov chain only allows the following typical transitions: S i,idle Convert to S i,hand S i,idle Convert to P i→jwalk or P i→jrun ; On the path, only the following typical transitions are allowed: P i→jwalk Convert to S j,idle P i→jrun Convert to S j,idle .

19. The multi-source fusion security early warning system according to claim 12, characterized in that, Estimating the inspection Markov chain and the transition matrix, including: If the staff state sequence {X} t If a chain follows a first-order discrete-time Markov chain, then: P r {X t+1 =x j ∣X t =x i ,X t-1 ,…}=P r {X t+1 =x j ∣X t =x i }=P ij , in: Let P be the transition probability matrix. ij For P r The probability of transitioning from state i to state j; Historical inspection routes Number of transfers: N ij =#{t∣X t =x i ,X t+1 =x j }, Then: The transition probability of the maximum likelihood estimation is: Where, N ij This is to count the number of transitions from i to j in history.

20. The multi-source fusion security early warning system according to claim 19, characterized in that, For the calculation formula of the transition probability of the maximum likelihood estimation, exponential weighted update is adopted: in, The transition probabilities are statistically obtained within a recent window, and λ controls the weight of new data. For the element P in the transition matrix of the next inspection after weighted update ij The value, The transition probability is the statistical result obtained within a recent window.

21. The multi-source fusion security early warning system according to claim 12, characterized in that, The calculation formula for the evaluation of the state risk is as follows: Among them, R total (t) is the comprehensive risk index, α, β, and γ are coefficients, and the sum of α, β, and γ is 1. lit and G lit These are the standard limits, E t and G t These represent the measured real-time electric field value and the real-time measured electric field gradient value, respectively, R. pred (t) represents the predicted risk score.

22. The multi-source fusion security early warning system according to claim 21, characterized in that, The determination formula for classifying the evaluated state risk is as follows: Where, 0<p1<p2, 0<λ1<λ2 are threshold parameters; Where, 0 represents safety, 1 represents potential danger, and 2 represents danger.

23. A computer device, characterized in that, Including: One or more processors; The processor is used to execute one or more programs; When the one or more programs are executed by the one or more processors, the method described in any one of claims 1-11 is implemented.

24. A computer-readable storage medium, characterized in that, There is a computer program stored thereon, and when the computer program is executed, the method described in any one of claims 1-11 is implemented.

Images