Power distribution network dispatching optimization method and device

Abstract

The invention relates to the technical field of power distribution network dispatching, in particular to a power distribution network dispatching optimization method and device. Constructing a dynamic digital twin of the power distribution network to be dispatched, and reversely correcting a line impedance parameter and an equipment efficiency coefficient in combination with a twin simulation result; multiple scheduling strategies are rehearsed in parallel, and the optimal scheduling strategy is selected through weighted scoring; generating a multi-energy flow collaborative scheduling instruction set according to the optimal scheduling strategy; dynamically distributing a control task by utilizing an edge node; and performing differentiated load response in combination with the power consumption behavior classification of the user. Space-time synchronous acquisition of electricity-heat-gas data is realized by deploying a multi-energy-flow sensor group, a multi-energy-flow collaborative scheduling instruction is generated by combining reverse parameter correction and scene rehearsal optimization of a dynamic digital twin, and dynamic resource allocation is realized based on an edge node dynamic allocation control task and load classification response. The new energy consumption rate capability is obviously improved, and the carbon emission intensity is reduced.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of power distribution network scheduling, and is a power distribution network scheduling optimization method and device. BACKGROUND

[0002] Power distribution network optimization scheduling is a core technical link of new power system construction, and its core task is to realize economic and efficient power supply by coordinating distributed power sources, energy storage systems, controllable loads and network equipment under the premise of meeting safety operation constraints. With the increasing demand for high proportion of renewable energy penetration and multi-energy interconnection, the scheduling object has expanded from the traditional power network to multi-energy flow coupled systems such as electricity, heat, cold and gas, and further needs to solve complex technical problems such as bilateral uncertainty of source and load, multi-energy complementary coordination and rapid fault self-healing.

[0003] However, the prior art has significant defects, including:

[0004] At the level of multi-energy flow data integration, the prior art mostly adopts an island-type monitoring scheme for independently collecting electric / thermal / gas data, and the electric, thermal and gas systems usually collect data independently, lacking unified time sequence alignment and abnormal cleaning mechanism, which makes it difficult for multi-source heterogeneous data to support joint optimization;

[0005] At the scheduling decision level, the prior art relies on static digital models with fixed parameters, which are difficult to reflect device aging and network topology changes, and the pre-rehearsal strategy ignores the transmission delay of the heat network and the dynamic constraints of the gas network pressure, causing the scheduling instructions to deviate from reality;

[0006] At the real-time control level, the communication delay of the centralized architecture restricts the response speed, and the fault handling relies on human experience, and the traditional load control adopts a "one-size-fits-all" mode, which cannot balance energy efficiency optimization and user comfort requirements.

[0007] These defects result in high wind and light abandonment rates, difficulty in reducing carbon emission intensity, and excessive time consumption for fault recovery. SUMMARY

[0008] The application provides a power distribution network scheduling optimization method and device, which overcomes the above-mentioned deficiencies of the prior art, and effectively solves the problem that the prior art mostly adopts an island-type monitoring scheme for independently collecting electric / thermal / gas data, lacking unified time sequence alignment and abnormal cleaning mechanism, which makes it difficult for multi-source heterogeneous data to support joint optimization.

[0009] One of the technical solutions of the application is achieved by the following measures: a power distribution network scheduling optimization method, comprising:

[0010] Real-time acquisition of multi-energy flow operation parameters of the to-be-scheduled power distribution network, and data synchronization and abnormal value cleaning based on a unified timestamp, to generate a time-space correlated multi-energy flow dataset, wherein the multi-energy flow operation parameters include node voltage, branch current, photovoltaic output, heat storage tank temperature, heat network flow and gas pipe network pressure;

[0011] Constructing a dynamic digital twin of the to-be-scheduled power distribution network, comparing the multi-energy flow dataset of the to-be-scheduled power distribution network with the twin simulation results of the dynamic digital twin, and when the voltage deviation is greater than 2% or the power flow direction is inconsistent, correcting the line impedance parameters and device efficiency coefficients in reverse based on device parameter correction rules, wherein the dynamic digital twin includes power grid topology, heat network pipeline transmission delay model and gas network node pressure balance equation;

[0012] Injecting a preset load fluctuation scenario and new energy output fluctuation scenario into the dynamic digital twin, and parallelly pre-playing multiple scheduling strategies, and selecting the optimal scheduling strategy through weighted scoring;

[0013] Generating a multi-energy flow coordinated scheduling instruction set according to the optimal scheduling strategy, wherein the multi-energy flow coordinated scheduling instruction set includes power instructions, heat instructions and gas instructions;

[0014] Dynamically allocating control tasks by using edge nodes;

[0015] Combining user electricity consumption behavior classification to perform differentiated load response.

[0016] The following is a further optimization or / and improvement of the above technical solutions:

[0017] The above-mentioned construction of the dynamic digital twin of the to-be-scheduled power distribution network includes:

[0018] Step S210, establishing a power grid-heat network-gas network coupling model, wherein the heat network pipeline transmission delay τ is as follows:

[0019]

[0020] Wherein, L is the pipeline length, and v is the hot water flow rate;

[0021] Step S220, constructing a gas network node pressure balance equation, and the principle formula is:

[0022] ΔP=ρgH+RQ 2 ;

[0023] Wherein, ΔP is the total pressure difference between the two ends of the pipeline, ρ is the gas density, g is the gravitational constant, H is the elevation difference, R is the pipeline resistance coefficient, and Q is the flow rate.

[0024] The above-mentioned device parameter correction rules include:

[0025] If |V实际 -V 仿真 If the value is greater than 0.02 pu, then update the corresponding transmission line resistance value.

[0026] If the heat power transfer error is >5%, then the heat pump COP value should be corrected.

[0027] Among them, V 实际 V is the effective value of the actual node voltage measured by the sensor. 仿真 The node voltage values ​​calculated for the dynamic digital twin, where R is the line resistance parameter to be calibrated. new These are the calibrated line resistance parameters.

[0028] The above-mentioned dynamic digital twin injects pre-set load fluctuation scenarios and new energy output fluctuation scenarios, performs parallel simulations of multiple scheduling strategies, and selects the optimal scheduling strategy through weighted scoring, including:

[0029] By using a dynamic digital twin of a pre-defined load fluctuation scenario and a new energy output fluctuation scenario, each scheduling strategy is simulated to obtain the corresponding economic index, environmental index, and safety margin.

[0030] The overall score of each scheduling strategy is obtained based on the weighted scoring. The scheduling strategy with the highest overall score is selected as the optimal scheduling strategy. The formula for calculating the overall score is as follows:

[0031] Score=0.4×θ1+0.3×θ2+0.3×θ3

[0032] Among them, θ1 is the economic index; θ2 is the environmental index; and θ3 is the safety margin.

[0033] The above-mentioned use of edge nodes to dynamically allocate control tasks includes:

[0034] A lightweight AI inference engine is embedded in the power distribution terminal firmware, which executes a localized Volt-Var optimization algorithm when the CPU utilization is less than or equal to 50%.

[0035] Q ref =K p ×(V ref -V meas )+K i ×∫(V ref -V meas )dt;

[0036] Among them, Q ref K is the setpoint for reactive power compensation. p and K i All are PID coefficients, V ref For voltage reference value, V measFor real-time measurement of voltage value, dt is the time integral operator;

[0037] When the CPU occupancy rate is greater than 50%, only the over-limit data is collected and determined according to the over-limit data judgment condition and uploaded, wherein the over-limit data judgment condition includes that the voltage is greater than 1.05pu or less than 0.95pu, the line load rate is greater than 95%, and the device temperature is greater than 85℃;

[0038] Switch to the preset safety mode when the communication is interrupted, and the switching process includes stopping energy storage charging, locking capacitor switching and maintaining the existing tap position.

[0039] The above combines the user's electricity consumption behavior classification to execute differentiated load response, including:

[0040] The industrial user automatically reduces the basic load by 5% when the system load rate is greater than 90%, which is implemented through the demand control system, and the response model corresponding to the demand control system is:

[0041] P adjust =α×P base ×ΔL;

[0042] Wherein, P adjust is the load power value to be adjusted, P base is the user's basic load power, α is the load elasticity coefficient, and ΔL is the system shortage ratio;

[0043] When the ambient temperature is higher than 30℃ and the voltage is lower than 0.95pu, the commercial user increases the air conditioning set temperature by 1℃, overwrites the building control system set value through Modbus-RTU protocol, and increases the dead zone control, that is, when the room temperature change rate is greater than 1℃ / min, the adjustment is suspended;

[0044] The resident user triggers the push energy-saving reminder at the peak period of electricity price, and the trigger condition is that the electricity price enters the peak period of time pricing and the load prediction curve rises by more than 10% compared with the baseline.

[0045] The above also includes responding to power distribution network fault events, performing fault detection and emergency recovery, including:

[0046] Establish a line temperature dynamic model:

[0047] T=T amb +I 2 R th ;

[0048] Wherein, T is the real-time temperature of the line conductor, T amb is the ambient temperature, I is the real-time current, and R th is the thermal resistance coefficient;

[0049] In response to a power distribution network fault event, the fault section is located within 50 milliseconds based on the current sudden change and voltage collapse, the island micro-grid is preferentially started to ensure power supply to key areas, and it is judged whether the real-time temperature T of the line conductor is greater than 90℃, if yes and lasts for 100 milliseconds, it is judged that there is a fire risk, the upstream and downstream breakers at the fault point are immediately disconnected, the dry powder fire extinguishing device is activated, and the fire situation is reviewed through optical fiber temperature measurement.

[0050] The above also includes feeding the load response execution amount, fault handling record and actual carbon emission intensity to the dynamic digital twin, using a model parameter evolution algorithm to drive device aging model updating and dispatching strategy iterative optimization, forming a closed-loop autonomous power distribution network operation control system, wherein the model parameter evolution algorithm includes:

[0051] Device aging factor update, the principle formula is: η new = η old × [1- β × (t / τ life ) 2 ]; wherein, η old is the initial efficiency of the device, η new is the updated efficiency of the device at the current time, the updated efficiency of the device at the current time is the aging acceleration coefficient, t is the cumulative running time of the device, τ life is the design life of the device;

[0052] Load elasticity coefficient calibration, when the actual reduction amount and the target deviation are greater than 10%, it is corrected according to the following formula:

[0053] α new = α old × (P real / P target )

[0054] Wherein, α old is the original load elasticity coefficient, α new is the corrected load elasticity coefficient, P real is the actual load reduction of the user, P target is the target reduction required by the dispatching system;

[0055] Every 24 hours, the dynamic digital twin is checked for full model: inject historical fault scenarios and verify the success rate of the reconstruction scheme.

[0056] The above also includes cross-system resilience collaboration, including:

[0057] Under the warning of extreme weather, the SOC of the energy storage is raised to 95%, the heat storage tank of the heat supply network is full, and the gas standby unit is started to run at no load;

[0058] After the failure, the multi-energy supply recovery is carried out, the photovoltaic-power storage-heat storage triple supply unit is preferentially started to support the key area, and the gas-electricity conversion equipment is gradually recovered;

[0059] A post-disaster evaluation matrix is established to record the power supply recovery rate, carbon emission reduction loss and equipment damage index, and the next disaster response strategy is optimized.

[0060] The second technical solution of the application is realized by the following measures: a power distribution network dispatching optimization device comprises:

[0061] A multi-energy flow data acquisition and synchronization unit acquires multi-energy flow operation parameters of the power distribution network to be dispatched in real time, synchronizes data based on a unified timestamp, and cleanses abnormal values to generate a multi-energy flow data set associated with time and space, wherein the multi-energy flow operation parameters include node voltage, branch current, photovoltaic output, heat storage tank temperature, heat network flow and gas pipe network pressure.

[0062] A digital twin modeling and dynamic correction unit constructs a dynamic digital twin of the power distribution network to be dispatched, compares the multi-energy flow data set of the power distribution network to be dispatched with the twin simulation results of the dynamic digital twin, and reversely corrects the line impedance parameters and device efficiency coefficients based on device parameter correction rules when the voltage deviation is greater than 2% or the power flow direction is inconsistent, wherein the dynamic digital twin comprises a power grid topology, a heat network pipe transmission delay model and a gas network node pressure balance equation.

[0063] A multi-strategy pre-rehearsal optimization unit injects preset load fluctuation scenarios and new energy output fluctuation scenarios into the dynamic digital twin, and pre-rehearsals multiple dispatch strategies in parallel, and selects the optimal dispatch strategy through weighted scoring.

[0064] A multi-energy flow coordinated dispatching instruction generation unit generates a multi-energy flow coordinated dispatching instruction set according to the optimal dispatch strategy, wherein the multi-energy flow coordinated dispatching instruction set comprises power instructions, heat instructions and gas instructions.

[0065] A dynamic task allocation unit dynamically allocates control tasks using edge nodes.

[0066] A load grading response unit executes differentiated load response in combination with user electricity behavior classification.

[0067] The following is a further optimization or / and improvement of the above-mentioned technical solutions of the application:

[0068] The above also includes:

[0069] A fault detection and emergency recovery unit responds to power distribution network fault events, performs fault detection and emergency recovery, including:

[0070] A line temperature dynamic model is established:

[0071] T=Tamb +I 2 R th ;

[0072] wherein, T is the real-time temperature of line conductor, T amb is the ambient temperature, I is the real-time current, R th is the thermal resistance coefficient;

[0073] In response to the power distribution network fault event, the fault section is located within 50 milliseconds based on the current mutation and voltage collapse, the island micro-grid is preferentially started to ensure power supply to key areas, and it is judged whether the real-time temperature T of the line conductor is greater than 90℃, if so and lasts for 100 milliseconds, then it is judged that there is a fire risk, the upstream and downstream breakers at the fault point are immediately disconnected, the dry powder fire extinguishing device is activated, and the fire situation is reviewed through optical fiber temperature measurement;

[0074] The closed-loop feedback and model evolution unit feeds back the load response execution amount, fault handling record and actual carbon emission intensity to the dynamic digital twin, adopts a model parameter evolution algorithm to drive the equipment aging model update and dispatching strategy iterative optimization, and forms a closed-loop autonomous power distribution network operation control system, wherein the model parameter evolution algorithm includes:

[0075] Equipment aging factor update, the principle formula is: η new = η old ×[1-β×(t / τ life ) 2 ]; wherein, η old is the initial efficiency of the equipment, η new is the updated efficiency of the equipment at the current time, the updated efficiency of the equipment at the current time is the aging acceleration coefficient, t is the cumulative running time of the equipment, τ life is the design life of the equipment;

[0076] Load elasticity coefficient calibration, when the actual reduction amount and the target deviation are greater than 10%, the following formula is used for correction;

[0077] α new = α old ×(P real / P target )

[0078] wherein, α old is the original load elasticity coefficient, α new is the corrected load elasticity coefficient, P real is the actual load reduction of the user, P target is the target reduction required by the dispatching system;

[0079] The dynamic digital twin is checked every 24 hours: inject historical fault scenarios to verify the success rate of the reconstruction scheme.

[0080] The cross-system resilience coordination unit comprises:

[0081] Under the extreme weather warning, the energy storage SOC is raised to 95%, the heat storage tank of the heat supply network is full, and the gas standby unit is started to run at no load;

[0082] After the failure, the multi-energy combined supply is restored, the photovoltaic-electricity storage-heat storage three-in-one unit is preferentially started to support the key area, and then the gas-electricity conversion equipment is gradually restored;

[0083] A post-disaster evaluation matrix is established to record the power supply recovery rate, carbon emission reduction loss and equipment damage index, and the next disaster response strategy is optimized.

[0084] The beneficial effects of the present application include:

[0085] Compared with the island monitoring scheme of the prior art which independently collects electric / heat / gas data, the present application has the shortcomings of difficulty in multi-source heterogeneous data fusion and lack of energy coupling constraints. The present application uses a cross-energy domain sensor group and a time-space synchronization mechanism to construct a time-space correlated multi-energy flow data set, realizes joint perception and collaborative optimization of the electric-thermal-gas operating state, and breaks through the technical bottleneck of fragmented scheduling of energy systems;

[0086] Compared with the static digital model of the prior art which relies on fixed parameters, the present application has the shortcomings of simulation distortion caused by equipment aging and deviation of scheduling decision from reality. The present application uses a dynamic digital twin and a reverse parameter correction mechanism to update the line impedance and equipment efficiency coefficient in real time through voltage / power flow deviation, provides a high-fidelity decision environment for multi-energy flow coordination, and ensures the credibility of multi-scenario rehearsal;

[0087] Compared with the prior art, the present application realizes parallel rehearsal of multiple scheduling strategies by using photovoltaic consumption / gas-electricity complementation / heat storage smoothing, dynamically selects the optimal scheduling strategy based on weighted scoring, realizes the Pareto optimal scheduling of the lowest light rejection rate, the minimum carbon emission and the most stable voltage, and solves the multi-objective decision-making contradiction;

[0088] Compared with the centralized cloud-edge collaborative architecture of the prior art, the present application has the disadvantage of communication time delay restricting the real-time control. The present application embodiment uses an edge node dynamic task allocation mechanism to adaptively switch between local optimization and key data upload mode according to CPU load, automatically enables a safety strategy when communication is interrupted, realizes the ability of millisecond-level reactive power control and fault isolation, and achieves the high flexibility control goal of "second-level response, millisecond protection";

[0089] Compared with the prior art of implementing "one-size-fits-all" load control, there are disadvantages of large industrial production loss, commercial comfort decline and low resident participation, the application adopts a classification response mechanism combined with user electricity consumption behavior, realizes precise mining and adjustment potential and guarantee of user experience through differentiated intervention of demand control / air conditioner temperature adjustment / APP push, and activates the value of demand side resources;

[0090] Compared with the prior art of relying on manual recovery of power supply after disaster, there are disadvantages of lack of power protection in key areas and low recovery efficiency, the application adopts a multi-level power protection and micro-grid rapid self-healing strategy, dynamically determines the fire risk combined with the line temperature model, realizes a proactive defense system of fault second-level positioning, preferential power supply and fire prevention linkage, and reconstructs a high-reliability and energy network. BRIEF DESCRIPTION OF DRAWINGS

[0091] FIG. 1 is a flowchart of a power distribution network scheduling optimization method provided by the application. Figure 1

[0092] FIG. 2 is a structure diagram of a power distribution network scheduling optimization device provided by the application. Figure 2

[0093] FIG. 3 is another structure diagram of a power distribution network scheduling optimization device provided by the application. Figure 3 DETAILED DESCRIPTION

[0094] The application is not limited by the following examples, and the specific implementation can be determined according to the technical scheme of the application and the actual situation.

[0095] Those skilled in the art can understand that, unless specifically stated, the "module" or "unit" in the embodiments of the application refers to a computer program or part of a computer program with a predetermined function, and works together with other related parts to achieve a predetermined target, and can be implemented entirely or partially by using software, hardware (such as processing circuit or memory) or combination thereof. Similarly, one processor (or multiple processors or memory) can be used to implement one or more modules or units. In addition, each module or unit can be part of an integral module or unit that includes the functions of the module or unit.

[0096] In addition, "multiple" in the embodiments of the application refers to two or more, and "first" and "second" are used to distinguish the description and cannot be understood as implying relative importance.

[0097] Based on this, the technical scheme of the application will be introduced and described below in combination with several examples.

[0098] Embodiment 1: As shown in the accompanying drawings, the embodiments of the application disclose a power distribution network scheduling optimization method, which comprises the following steps: Figure 1 Embodiment 1: As shown in the accompanying drawings, the embodiments of the application disclose a power distribution network scheduling optimization method, which comprises the following steps:​​​

[0099] In step S110, the multi-energy flow operation parameters of the power distribution network to be dispatched are collected in real time, and data synchronization and abnormal value cleaning are performed based on a unified timestamp to generate a time-space correlated multi-energy flow data set, wherein the multi-energy flow operation parameters include node voltage, branch current, photovoltaic output, heat tank temperature, heat network flow, and gas pipe network pressure.

[0100] In this step, the multi-energy flow operation parameters of the power distribution network to be dispatched are collected in real time by deploying a sensor group in the power, heat, and gas networks. After collection, a hierarchical compression transmission mechanism is used, including:

[0101] (1) The power data is transmitted at an interval of 1 second through the GOOSE protocol and is compressed by Huffman coding;

[0102] (2) The heat network and gas network data are transmitted at an interval of 5 seconds through the MQTT protocol, and differential encoding is used to reduce redundancy;

[0103] (3) The multi-source data packets are fused at the edge node, and the transmission delay difference is eliminated by a time calibration algorithm.

[0104] The hierarchical compression transmission mechanism (GOOSE protocol + Huffman coding for power data, MQTT protocol + differential encoding for heat / gas network data) described above is used to fuse multi-source data and calibrate time at the edge node, effectively reducing the communication bandwidth occupation, eliminating the timing misalignment problem of cross-energy data transmission, and ensuring the real-time and consistency of the data for multi-energy flow joint optimization.

[0105] Compared with the existing island-type monitoring scheme that independently collects electric / thermal / gas data, the method has the disadvantages of difficulty in fusing multi-source heterogeneous data and lack of energy coupling constraints. The step uses a cross-energy domain sensor group and a time-space synchronization mechanism to construct a time-space correlated multi-energy flow data set, realizes joint perception and collaborative optimization of the electric-thermal-gas operation state, and breaks through the technical bottleneck of fragmented energy system scheduling.

[0106] In step S120, a dynamic digital twin of the power distribution network to be dispatched is constructed. The multi-energy flow data set of the power distribution network to be dispatched is compared with the simulation results of the dynamic digital twin. When the voltage deviation is greater than 2% or the power flow direction is inconsistent, the line impedance parameters and device efficiency coefficients are corrected in reverse based on device parameter correction rules, wherein the dynamic digital twin includes the power grid topology structure, the heat network pipeline transmission delay model, and the gas network node pressure balance equation.

[0107] In step S130, a preset load fluctuation scenario and a new energy output fluctuation scenario are injected into the dynamic digital twin, and multiple scheduling strategies are pre- simulated in parallel. The optimal scheduling strategy is selected by weighted scoring.

[0108] This step specifically includes:

[0109] (1) using the dynamic digital twin injected with the preset load fluctuation scenario and the new energy output fluctuation scenario, each scheduling strategy is preformed, and corresponding economic index, environmental protection index and safety margin are obtained, wherein the economic index is the reciprocal of the total operation cost, the environmental protection index is negatively correlated with the carbon emission intensity, and the safety margin is calculated according to the minimum voltage deviation rate and the line load rate;

[0110] (2) based on the weighted score, the comprehensive score of each scheduling strategy is obtained, and the scheduling strategy with the highest comprehensive score is taken as the optimal scheduling strategy, wherein the comprehensive score calculation formula is as follows:

[0111] Score=0.4×θ1+0.3×θ2+0.3×θ3

[0112] Wherein, θ1 is the economic index; θ2 is the environmental protection index; and θ3 is the safety margin.

[0113] Step S140: generating a multi-energy flow collaborative scheduling instruction set according to the optimal scheduling strategy, wherein the multi-energy flow collaborative scheduling instruction set includes power instructions, heat instructions and gas instructions;

[0114] In this step, the power instruction controls the photovoltaic inverter limited emission ratio and the energy storage charge-discharge curve, the heat instruction specifies the heat pump power adjustment amplitude and the heat storage tank heat release timing, and the gas instruction sets the fuel cell start-stop state and the pipe network pressure threshold.

[0115] Step S150: using the edge node to dynamically allocate control tasks.

[0116] Step S160: executing differentiated load response in combination with user power consumption behavior classification.

[0117] The embodiment of the application discloses a power distribution network scheduling optimization method, through the sensor group deployed in the power, heat and gas network, the space-time synchronous collection of electric-thermal-gas operation parameters is realized, the reverse parameter correction and scheduling strategy pre-performance are realized in combination with the dynamic digital twin, the optimal scheduling strategy is obtained, the corresponding multi-energy flow collaborative scheduling instruction is generated, the millisecond reactive power optimization is realized by using the edge node to dynamically allocate tasks, the differentiated load response is executed in combination with the user behavior classification, the second-level fault self-healing is realized, and finally the scheduling optimization control system from data perception, decision optimization to execution is formed, and the power distribution network is driven to evolve in the direction of safety, low carbon, flexibility and intelligence.

[0118] Embodiment 2: The embodiment of the application is a further optimization of the above-mentioned embodiment, in particular,

[0119] The dynamic digital twin of the power distribution network to be scheduled is constructed, including:

[0120] Step S210, a power grid-heat grid-gas grid coupling model is established, wherein the heat grid pipeline transmission delay tau is as follows:

[0121]

[0122] Wherein, L is the pipeline length, and v is the hot water flow rate;

[0123] Step S220, a gas grid node pressure balance equation is constructed, and the principle formula is as follows:

[0124] Delta P is the total pressure difference of the pipeline, rho is the gas density, g is the gravitational constant, H is the elevation difference, R is the pipeline resistance coefficient, and Q is the flow. 2 ;

[0125] Wherein, Delta P is the total pressure difference of the pipeline, rho is the gas density, g is the gravitational constant, H is the elevation difference, R is the pipeline resistance coefficient, and Q is the flow.

[0126] The device parameter correction rule comprises:

[0127] If |V 实际 -V 仿真 |>0.02pu, the corresponding power transmission line resistance value is updated

[0128] If the heat power transmission error is greater than 5%, the heat pump COP value is corrected.

[0129] Wherein, V 实际 is the actual voltage effective value of the node measured by the sensor, V 仿真 is the node voltage value calculated by the dynamic digital twin, R is the line resistance parameter to be calibrated, and R new is the calibrated line resistance parameter.

[0130] Compared with the static digital model of the prior art relying on fixed parameters, there are disadvantages that equipment aging leads to simulation distortion and scheduling decision deviation from the actual situation, and the embodiment of the present application adopts a dynamic digital twin and a reverse parameter correction mechanism, which updates the line impedance and the equipment efficiency coefficient in real time through voltage / power flow deviation, provides a high-fidelity decision environment for multi-energy flow cooperation, and guarantees the credibility of multi-scenario rehearsal.

[0131] Embodiment 3: The embodiment of the present application is a further optimization of the above-mentioned embodiment, wherein the plurality of scheduling strategies comprise:

[0132] (1) photovoltaic preferential consumption mode, forcibly closing the gas peak shaving unit, absorbing the heat energy converted by the excess photovoltaic electric energy through the heat storage tank;

[0133] (2) gas-electricity complementary peak shaving mode, starting the gas unit to supply power during the peak period of electricity price, and balancing the power grid load by using the electric hydrogen production equipment;

[0134] (3) heat storage stabilizes fluctuation mode, when the photovoltaic power fluctuation rate exceeds 15% / minute, the heat storage tank is charged and discharged to offset the heat network load disturbance.

[0135] The embodiment of the application adopts photovoltaic accommodation / gas-electricity complementation / heat storage stabilization to realize parallel pre-rehearsal of multiple scheduling strategies, dynamically selects the optimal scheduling strategy based on weighted scoring, realizes Pareto optimal scheduling with the lowest light rejection rate, the minimum carbon emission and the most stable voltage, and solves the contradiction of multi-objective decision-making.

[0136] Embodiment 4: The embodiment of the application is a further optimization of the above-mentioned embodiments, wherein the edge node is used to dynamically allocate control tasks, including:

[0137] Step S310, a lightweight AI inference engine is embedded in the power distribution terminal firmware, and a localized Volt-Var optimization algorithm is executed when the CPU occupancy rate is less than or equal to 50%:

[0138] Q ref = K p × (V ref -V meas ) + K i × ∫ (V ref -V meas ) dt

[0139] Wherein, Q ref is the reactive power compensation set value, K p and K i are PID coefficients, V ref is the voltage reference value, V meas is the real-time measured voltage value, and dt is the time integral operator.

[0140] Step S320, when the CPU occupancy rate is greater than 50%, only collect and determine and upload the overrun data according to the overrun data judgment condition, wherein the overrun data judgment condition includes that the voltage is greater than 1.05pu or less than 0.95pu, the line load rate is greater than 95%, and the device temperature is greater than 85℃;

[0141] Step S330, switch to a preset safety mode when communication is interrupted, and the switching process includes stopping energy storage charging, locking capacitor switching and maintaining the existing tap position.

[0142] Compared with the centralized cloud edge collaborative architecture of the prior art, there is a disadvantage that the communication time delay restricts the real-time control, the embodiment of the application adopts an edge node dynamic task allocation mechanism, adaptively switches between local optimization and key data upload mode according to CPU load, automatically enables a safety strategy when communication is interrupted, realizes the ability of millisecond-level reactive power control and fault isolation, and achieves the high flexibility control goal of "second-level response and millisecond protection".

[0143] Example 5: This embodiment of the invention is a further optimization of the above embodiments, wherein differentiated load response is performed based on user electricity consumption behavior classification, including:

[0144] Step S410: When the system load rate is greater than 90%, the industrial user automatically reduces the base load by 5%, specifically implemented through the demand control system. The response model of the demand control system is as follows:

[0145] P adjust =α×P base ×ΔL;

[0146] Among them, P adjust P is the load power value that needs to be adjusted. base Where α is the user's basic load power, α is the load flexibility coefficient, and ΔL is the system deficit ratio;

[0147] Step S420: When the ambient temperature is higher than 30℃ and the voltage is lower than 0.95pu, the commercial user raises the air conditioner set temperature by 1℃, rewrites the building control system set value through the Modbus-RTU protocol, and adds dead zone control, that is, when the room temperature change rate is greater than 1℃ / minute, the adjustment is paused.

[0148] Step S430: Residential users trigger energy-saving reminders during peak electricity price periods. The triggering conditions are that the electricity price enters the peak period of time-of-use pricing and the load forecast curve rises by more than 10% from the baseline.

[0149] It should be noted that triggering a push notification for energy saving can trigger a push notification for energy saving from a related app.

[0150] Compared to the existing technology that implements "one-size-fits-all" load control, which has drawbacks such as large losses from industrial shutdowns, reduced commercial comfort, and low resident participation, the embodiments of this invention adopt a classification response mechanism that combines user electricity consumption behavior. Through differentiated interventions such as demand control, air conditioning temperature adjustment, and APP push notifications, it achieves a dual breakthrough of accurately tapping into regulation potential and ensuring user experience, thereby activating the value of demand-side resources.

[0151] Example 6: This embodiment of the invention is a further optimization of the above embodiments, and also includes responding to distribution network fault events and performing fault detection and emergency recovery, including:

[0152] Step S510, establish a dynamic model of line temperature:

[0153] T = T amb +I 2 R th ;

[0154] Where T is the real-time temperature of the line conductor, T amb R represents ambient temperature, I represents real-time current, and R represents ambient temperature. th It is the thermal resistance coefficient;

[0155] Step S520: In response to the power distribution network fault event, locate the fault section based on the current surge and voltage collapse within 50 milliseconds, prioritize the activation of the islanded microgrid to ensure power supply to key areas, and determine whether the real-time temperature T of the line conductor is greater than 90℃. If it is and lasts for 100 milliseconds, it is determined that there is a fire risk, and immediately disconnect the upstream and downstream circuit breakers of the fault point, activate the dry powder fire extinguishing device, and verify the fire situation through fiber optic temperature measurement.

[0156] Among them, a multi-level priority mechanism is adopted to ensure power supply in key areas, including: hospitals and emergency command centers are at level 1 (power outages are not allowed under any circumstances), and schools and transportation hubs are at level 2.

[0157] This invention achieves millisecond-level fault location and fire risk warning through a dynamic line temperature model. Combined with a multi-level priority mechanism, it ensures power supply to key areas and rapid load shedding of microgrids, reducing fault recovery time from 5 minutes to less than 500ms, while effectively preventing electrical fire accidents.

[0158] Example 7: This embodiment of the invention is a further optimization of the above embodiments. After implementing differentiated load response based on user electricity consumption behavior classification, it also includes a feedback optimization process, specifically including:

[0159] Load response execution data, fault handling records, and actual carbon emission intensity are fed back to a dynamic digital twin. A model parameter evolution algorithm drives equipment aging model updates and scheduling strategy iterative optimization, forming a closed-loop autonomous distribution network operation control system. The model parameter evolution algorithm includes:

[0160] (1) Equipment aging factor renewal, the principle formula is: η new =η old ×[1-β×(t / τ life ) 2 ]; where η old η is the initial efficiency of the equipment. new The efficiency of the equipment after the update at the current moment is the aging acceleration factor, t is the cumulative running time of the equipment, and τ is the aging acceleration factor. life The design life of the equipment;

[0161] (2) Load elasticity coefficient calibration: when the actual reduction amount deviates from the target by more than 10%, it shall be corrected according to the following formula;

[0162] α new =α old ×(P real / P target )

[0163] Where, α old α is the original load elasticity coefficient.new P is the corrected load elasticity coefficient real P is the actual load reduction of the user target P is the target reduction required by the dispatching system;

[0164] (3) Full model verification of dynamic digital twin every 24 hours: inject historical fault scenarios and verify the success rate of reconstruction scheme execution.

[0165] Example 8: The embodiment of the present application is a further optimization of the above-mentioned embodiment, which further includes cross-system resilience collaboration, including:

[0166] (1) Under the warning of extreme weather, the SOC of energy storage is raised to 95%, the thermal network storage tank is full, and the gas standby unit is started to run at no load;

[0167] (2) After the failure, the multi-energy combined supply is restored, and the photovoltaic-electricity storage-heat storage three-in-one unit is preferentially started to support key areas, and then the gas-electricity conversion equipment is gradually restored;

[0168] (3) Establish a post-disaster evaluation matrix to record the power supply recovery rate, carbon emission loss and equipment damage index, and optimize the next disaster response strategy.

[0169] The embodiment of the present application implements cross-system collaborative defense of energy storage SOC raising, thermal network full storage and gas unit pre-starting under extreme disasters, combined with the post-disaster photovoltaic-electricity storage-heat storage three-in-one recovery strategy, to improve the power supply recovery rate and reduce carbon emission loss.

[0170] Example 9: Use the power distribution network dispatching optimization method disclosed in the present application to realize multi-energy collaborative dispatching in summer peak

[0171] Scenario background: A 30MW photovoltaic power station is connected to the power distribution network of an industrial park, the thermal network supplies 500,000 square meters of factory refrigeration, and the gas network drives two 10MW gas turbine units. The summer noon temperature is 38℃, the industrial load increases suddenly, the photovoltaic output fluctuation rate exceeds 20% / minute, and the power grid voltage fluctuation reaches ±8%.

[0172] I. Multi-energy flow data synchronization:

[0173] The power sensor collects node voltage (fluctuates to 0.92-1.07pu) at 1 second intervals and transmits it through GOOSE protocol compression; the thermal network flow meter sends chilled water flow data at 5 second intervals, using differential encoding to reduce temperature redundancy; the gas pressure sensor monitors the pipeline pressure (1.2-2.0MPa) in real time. The edge node fuses the three types of data, and the timestamp deviation is calibrated to be less than 10ms.

[0174] II. Reverse correction based on dynamic digital twin:

[0175] A dynamic digital twin is constructed including a 22 km steam pipeline (time delay τ = 300 s) and gas network node pressure balance equations. The simulated value of the 110 kV bus voltage (1.04 pu) deviates from the measured value (1.01 pu) by more than 2%, and the line impedance parameter R is corrected in the opposite direction: R = 0.015 × (1.01 / 1.04) ≈ 0.0146 Ω, and the heat pump COP value is corrected from 3.2 to 3.0 due to the decrease in heat exchange efficiency. new

[0176] III. Multiple scheduling strategy pre-rehearsal optimization

[0177] Inject photovoltaic fluctuation ± 25% scene, pre-rehearsal results:

[0178] Photovoltaic priority mode: turn off gas units, the remaining electricity drives the heat pump to refrigerate, and the heat storage tank temperature rises from 5°C to 12°C;

[0179] Gas-electricity complementary mode: start gas units for peak shaving, and use excess electricity to produce hydrogen;

[0180] Heat storage smoothing mode: heat storage tank absorbs 40% of photovoltaic fluctuation.

[0181] Weighted score selects heat storage smoothing mode.

[0182] IV. Dynamic allocation of edge tasks

[0183] Feeder terminal CPU load 45%, execute local Volt-Var control as follows:

[0184] Q ref = 2.5 × (1.02 - V_meas) + 0.8 × ∫(1.02 - V_meas)dt

[0185] Raise the voltage to 1.00 ± 0.01 pu.

[0186] V. Differentiated load response

[0187] System load rate 92%, industrial users automatically reduce 5% load (α = 0.9), response model as follows:

[0188] P adjust = α × P base × ΔL = 0.9 × 800 kW × 0.05 = 36 kW

[0189] Commercial user air conditioning temperature up 1°C (room temperature 28→29°C);

[0190] Residential area APP pushes "electricity price peak" reminder.

[0191] VI. Fault linkage treatment

[0192] ​Photovoltaic feeder current surge 200%, located as DC arc fault. Line temperature rises to 95℃ within 1 second, forced outage of fault area, start of micro-grid power supply for hospital (1st level power protection). Tie-in switch 500ms within the reconstructed power grid restores 95% load.

[0193] Seven, closed-loop parameter optimization

[0194] Actual industrial load reduction of 32kW (target 36kW), update of elastic coefficient, as follows:

[0195] α new =α old ×(P real / P target )=0.9×(32 / 36)=0.8

[0196] The efficiency after calibration of the equipment aging model is as follows:

[0197] η new =η old ×[1-β×(t / τ life ) 2 ]=0.93×[1-0.02×(8 / 30) 2 ]=0.914

[0198] Based on the above, the light rejection rate is reduced to 4%, and the peak load is reduced by 8%, and the fault recovery time is <0.5 seconds.

[0199] Example 10: Use of the disclosed power distribution network dispatching optimization method to achieve seasonal ice and snow disaster resilience recovery

[0200] Scenario background: cold invasion causes power transmission icing, 220kV main supply line trips, and the power distribution network operates in island mode. Key protection of hospital (1st level) and transportation hub (2nd level) power supply, gas source pressure shortage restricts gas turbine start-stop.

[0201] Technical solution implementation:

[0202] I. Multi-energy flow emergency monitoring:

[0203] Optical fiber temperature measurement shows that the conductor icing thickness is 12mm; the gas network pressure drops to 0.8MPa (critical gas supply threshold); the heat network return water temperature drops to 40℃.

[0204] II. Disaster simulation and dispatching:

[0205] Digital twin injects "main network power failure" scenario:

[0206] Start cross-system resilience collaboration: energy storage SOC from 60% to 95% emergency charging, heat storage tank full (95℃), gas turbine preheating at no load;

[0207] Optimization strategy: priority to enable hospital roof photovoltaic (500 kW) + energy storage (2 MWh) + heat storage tank (300 kWh) tri-generation unit.

[0208] III. Edge intelligence control:

[0209] When the communication is interrupted, the DTU automatically switches to the safety mode: stop charging the energy storage, lock the capacitor switching, and maintain the OLTC tap position.

[0210] IV. Multi-level power protection and heat network linkage:

[0211] The hospital microgrid switches to island operation, and the gas boiler switches to the heat storage tank for heating; the traffic hub (2 levels) load is reduced by 30% to maintain pressure.

[0212] V. Fault active defense:

[0213] Ice-coated line current surge, temperature model warning: T = -5℃ + (650A) 2 ×0.005 = 82℃, continuously exceeding 90℃ for 80ms, starting the ice melting device and removing non-critical load.

[0214] VI. Post-disaster evaluation optimization:

[0215] Record power supply recovery rate (1st level 100%, 2nd level 85%), equipment damage index (conductor sag increased by 20%), carbon loss (diesel engine caused carbon emissions +18%), and generate resilience optimization matrix.

[0216] VII. Closed-loop model evolution

[0217] Twin body verification ice-coated scene reconstruction scheme, success rate 99%; load elasticity coefficient α is corrected from 0.7 to 0.65 according to response deviation.

[0218] In summary, the hospital has 100% continuous power supply, the key load recovery time is less than 45 seconds, and the disaster dispatching strategy iteration efficiency is improved by 30%.

[0219] Embodiment 9: as shown in the accompanying Figure 2 The power distribution network dispatching optimization device disclosed by the embodiment of the application comprises:

[0220] A multi-energy flow data acquisition and synchronization unit acquires multi-energy flow operation parameters of a to-be-dispatched power distribution network in real time, synchronizes data and cleanses abnormal values based on a unified timestamp, and generates a multi-energy flow data set associated with time and space, wherein the multi-energy flow operation parameters comprise node voltage, branch current, photovoltaic output, heat storage tank temperature, heat network flow, and gas pipe network pressure.

[0221] The digital twin modeling and dynamic correction unit constructs a dynamic digital twin of the to-be-dispatched power distribution network, compares the multi-energy flow data set of the to-be-dispatched power distribution network with a twin simulation result of the dynamic digital twin, and reversely corrects line impedance parameters and device efficiency coefficients based on device parameter correction rules when a voltage deviation is greater than 2% or a power flow direction is inconsistent, wherein the dynamic digital twin includes a power grid topology structure, a heat network pipeline transmission delay model, and a gas network node pressure balance equation.

[0222] The multi-strategy pre-rehearsal optimization unit injects preset load fluctuation scenarios and new energy output fluctuation scenarios into the dynamic digital twin, pre-rehearsals multiple dispatch strategies in parallel, and selects an optimal dispatch strategy through weighted scoring.

[0223] The multi-energy flow coordinated dispatch instruction generation unit generates a multi-energy flow coordinated dispatch instruction set according to the optimal dispatch strategy, wherein the multi-energy flow coordinated dispatch instruction set includes power instructions, heat instructions, and gas instructions.

[0224] The dynamic task allocation unit dynamically allocates control tasks by using edge nodes.

[0225] The load grading response unit executes differentiated load responses in combination with user electricity consumption behavior classification.

[0226] Embodiment 10: As shown in the accompanying Figure 3 The power distribution network dispatching optimization device disclosed by the embodiment of the present application comprises:

[0227] The multi-energy flow data acquisition and synchronization unit acquires multi-energy flow operation parameters of the to-be-dispatched power distribution network in real time, performs data synchronization and outlier cleaning based on a unified timestamp, and generates a time-space correlated multi-energy flow data set, wherein the multi-energy flow operation parameters include node voltage, branch current, photovoltaic output, heat storage tank temperature, heat network flow, and gas pipe network pressure.

[0228] The digital twin modeling and dynamic correction unit constructs a dynamic digital twin of the to-be-dispatched power distribution network, compares the multi-energy flow data set of the to-be-dispatched power distribution network with a twin simulation result of the dynamic digital twin, and reversely corrects line impedance parameters and device efficiency coefficients based on device parameter correction rules when a voltage deviation is greater than 2% or a power flow direction is inconsistent, wherein the dynamic digital twin includes a power grid topology structure, a heat network pipeline transmission delay model, and a gas network node pressure balance equation.

[0229] The multi-strategy pre-rehearsal optimization unit injects preset load fluctuation scenarios and new energy output fluctuation scenarios into the dynamic digital twin, pre-rehearsals multiple dispatch strategies in parallel, and selects an optimal dispatch strategy through weighted scoring.

[0230] The multi-energy flow cooperative scheduling instruction generation unit generates a multi-energy flow cooperative scheduling instruction set according to the optimal scheduling strategy, wherein the multi-energy flow cooperative scheduling instruction set includes power instructions, heat instructions, and gas instructions.

[0231] The dynamic task allocation unit dynamically allocates control tasks by using the edge node.

[0232] The load grading response unit executes differentiated load response in combination with user electricity consumption behavior classification.

[0233] The fault detection and emergency recovery unit responds to power distribution network fault events, executes fault detection and emergency recovery, including:

[0234] A line temperature dynamic model is established:

[0235] T = T amb + I 2 R th ;

[0236] Wherein, T is the real-time temperature of the line conductor, T amb is the ambient temperature, I is the real-time current, and R th is the thermal resistance coefficient;

[0237] In response to a power distribution network fault event, the fault section is located based on the current mutation and voltage collapse within 50 milliseconds, the island micro-grid is started to ensure power supply to key areas, and it is judged whether the real-time temperature T of the line conductor is greater than 90℃. If yes and lasts for 100 milliseconds, it is determined that there is a fire risk, the upstream and downstream breakers at the fault point are immediately disconnected, the dry powder fire extinguishing device is activated, and the fire situation is reviewed through optical fiber temperature measurement.

[0238] The closed-loop feedback and model evolution unit feeds back the load response execution amount, fault handling record, and actual carbon emission intensity to the dynamic digital twin, updates the equipment aging model and iteratively optimizes the scheduling strategy by using a model parameter evolution algorithm, and forms a closed-loop autonomous power distribution network operation control system. The model parameter evolution algorithm includes:

[0239] The equipment aging factor is updated, and the principle formula is: η new = η old × [1- β × (t / τ life ) 2 ]; wherein η old is the initial efficiency of the equipment, η new is the updated efficiency of the equipment at the current time, the updated efficiency of the equipment at the current time is the aging acceleration coefficient, t is the cumulative running time of the equipment, and τ life is the design life of the equipment.

[0240] The load elasticity coefficient is calibrated, and when the actual reduction amount and the target deviation are greater than 10%, it is corrected according to the following formula:

[0241] α new = α old × (P real / P target )

[0242] Wherein, α old is the original load elasticity coefficient, α new is the corrected load elasticity coefficient, P real is the actual load reduction of the user, P target is the target reduction required by the dispatching system.

[0243] Full model verification of dynamic digital twin every 24 hours: inject historical fault scenarios and verify the success rate of reconstruction scheme execution.

[0244] Cross-system resilience coordination unit, including:

[0245] Under the warning of extreme weather, the SOC of energy storage is raised to 95%, the heat storage tank of heat supply network is full, and the gas standby unit is started to run at no load;

[0246] After the failure, the multi-energy combined supply is restored, the photovoltaic-electricity storage-heat storage three-in-one unit is preferentially started to support key areas, and then the gas-electricity conversion equipment is gradually restored;

[0247] Establish a post-disaster evaluation matrix to record the power restoration rate, carbon emission reduction loss and equipment damage index, and optimize the response strategy for the next disaster.

[0248] The above is only a specific embodiment of the present application, which has strong adaptability and implementation effect, but the protection scope of the present application is not limited thereto, any skilled person in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application, therefore, the equivalent changes made according to the claims of the present application still belong to the scope covered by the present application.

Claims

1. A power distribution network dispatch optimization method, characterized in that, The method comprises the following steps: Real-time acquisition of multi-energy flow operation parameters of the to-be-scheduled power distribution network, and data synchronization and abnormal value cleaning based on a unified timestamp to generate a time-space correlated multi-energy flow data set, wherein the multi-energy flow operation parameters include node voltage, branch current, photovoltaic output, heat storage tank temperature, heat network flow and gas pipe network pressure; Constructing a dynamic digital twin of the to-be-scheduled power distribution network, comparing the multi-energy flow data set of the to-be-scheduled power distribution network with the simulation results of the digital twin, and when the voltage deviation is greater than 2% or the power flow direction is inconsistent, correcting the line impedance parameters and device efficiency coefficients in the reverse direction based on the device parameter correction rules, wherein the dynamic digital twin includes the power grid topology, the heat network pipe transmission delay model and the gas network node pressure balance equation; Injecting a preset load fluctuation scenario and a new energy output fluctuation scenario into the dynamic digital twin, and pre-rehearsing multiple scheduling strategies in parallel, and selecting the optimal scheduling strategy through weighted scoring; Generating a multi-energy flow coordinated scheduling instruction set according to the optimal scheduling strategy, wherein the multi-energy flow coordinated scheduling instruction set includes power instructions, heat instructions and gas instructions; Dynamically allocating control tasks by using edge nodes; Executing differentiated load response in combination with user electricity consumption behavior classification.

2. The power distribution grid dispatch optimization method of claim 1, wherein, Constructing a dynamic digital twin of the to-be-scheduled power distribution network, comprising: Step S210, establishing an electric grid-heat network-gas network coupling model, wherein the heat network pipe transmission delay τ is as follows: Wherein, L is the pipe length, and v is the hot water flow rate; Step S220, constructing a gas network node pressure balance equation, and the principle formula is as follows: ΔP = pgH + RQ 2 ; Wherein, ΔP is the total pressure difference at both ends of the pipe, ρ is the gas density, g is the gravitational constant, H is the elevation difference, R is the pipe resistance coefficient, and Q is the flow rate; Or / and, The device parameter correction rule comprises: If |V 实际 -V 仿真 If |V > 0.02pu, update the corresponding transmission line resistance value If the heat power transmission error is >5%, the heat pump COP value is corrected; V 实际 is the actual voltage effective value of the node measured by the sensor, V 仿真 is the node voltage value calculated by the dynamic digital twin, R is the line resistance parameter to be calibrated, R new is the calibrated line resistance parameter.

3. The power distribution network dispatch optimization method of claim 1 or 2, wherein, Injecting a preset load fluctuation scenario and a new energy output fluctuation scenario into the dynamic digital twin, and pre-rehearsing multiple scheduling strategies in parallel, and selecting the optimal scheduling strategy through weighted scoring, comprising: Using the dynamic digital twin injected with the preset load fluctuation scenario and the new energy output fluctuation scenario to pre-rehearse each scheduling strategy to obtain corresponding economic index, environmental protection index and safety margin; Based on the weighted scoring, the comprehensive score of each scheduling strategy is obtained, and the scheduling strategy with the highest comprehensive score is selected as the optimal scheduling strategy, wherein the comprehensive score calculation formula is as follows: Score = 0.4 × θ1 + 0.3 × θ2 + 0.3 × θ3 Wherein, θ1 is the economic index; θ2 is the environmental protection index; and θ3 is the safety margin.

4. The power distribution network dispatch optimization method of claim 1 or 2, wherein, Dynamically allocating control tasks by using edge nodes, comprising: Embedding a lightweight AI inference engine in the power distribution terminal firmware, and when the CPU occupancy rate is less than or equal to 50%, executing a localized Volt-Var optimization algorithm: Q ref = K p × (V ref - V meas ) + K i × ∫(V ref - V meas ) dt; wherein Q ref is a reactive power compensation set value, K p and K i are PID coefficients, V ref is a voltage reference value, V meas is a real-time measured voltage value, and dt is a time integral operator; When the CPU occupancy rate is greater than 50%, only collect and determine and upload over-limit data according to over-limit data judgment conditions, wherein the over-limit data judgment conditions include voltage greater than 1.05pu or less than 0.95pu, line load rate greater than 95%, and device temperature greater than >85℃. Switch to preset safety mode when communication is interrupted, the switching process includes stopping energy storage charging, locking capacitor switching and maintaining the existing tap position.

5. The power distribution network dispatch optimization method of claim 1 or 2, wherein, Differential load response is performed in combination with user electricity consumption behavior classification, including: Industrial users automatically reduce 5% of the basic load when the system load rate is greater than 90%, which is implemented through a demand control system, and the corresponding response model of the demand control system is: P adjust = a x P base x AL; Wherein, P adjust is the load power value to be adjusted, P base is the user base load power, a is the load elasticity coefficient, and ΔL is the system shortage ratio. Commercial users increase the air conditioning set temperature by 1℃ when the ambient temperature is higher than 30℃ and the voltage is lower than 0.95pu, which is implemented by rewriting the building control system set value through Modbus-RTU protocol and adding dead zone control, i.e. when the room temperature change rate is greater than 1℃ / min, the adjustment is suspended; Residential users trigger a push energy-saving reminder during peak electricity pricing periods when the electricity price enters the peak period of time-based pricing and the load forecast curve rises by more than 10% compared to the baseline.

6. The power distribution network dispatch optimization method of claim 1 or 2, wherein, It also includes responding to power distribution network fault events, performing fault detection and emergency recovery, including: Establishing a line temperature dynamic model: T = T amb + I 2 R th ; Wherein, T is the real-time temperature of the line conductor, T amb is the ambient temperature, I is the real-time current, R th is the thermal resistance coefficient; In response to power distribution network fault events, locate the fault section within 50 milliseconds based on current sudden change and voltage collapse, preferentially start the island microgrid to ensure power supply to key areas, and determine whether the line conductor real-time temperature T is greater than 90℃, if so and lasts for 100 milliseconds, then determine that there is a fire risk, immediately disconnect the upstream and downstream breakers at the fault point, activate the dry powder fire extinguishing device, and review the fire situation through fiber-optic temperature measurement.

7. The power distribution network dispatch optimization method of claim 1 or 2, wherein, It also includes feeding back the load response execution amount, fault handling records, and actual carbon emission intensity to the dynamic digital twin, using a model parameter evolution algorithm to drive equipment aging model updates and dispatching strategy iterative optimization, forming a closed-loop autonomous power distribution network operation control system, wherein the model parameter evolution algorithm includes: The principle formula for updating the device aging factor is: η new = η old × [1-β×(t / τ life ) 2 ]; wherein, η old is the initial efficiency of the device, η new is the updated efficiency of the device at the current time, the updated efficiency of the device at the current time is the aging acceleration coefficient, t is the cumulative running time of the device, τ life is the design life of the device; Load elasticity coefficient calibration, when the actual reduction amount deviates from the target by more than 10%, correct it according to the following formula; α new = α old × (P real / P target ) wherein, α old is the original load elasticity coefficient, α new is the corrected load elasticity coefficient, P real is the actual load reduction of the user, P target is the target reduction required by the dispatching system; Every 24 hours, perform full model verification on the dynamic digital twin: inject historical fault scenarios and verify the success rate of the reconstruction scheme.

8. The power distribution network dispatch optimization method of any one of claims 1 to 7, wherein, It also includes cross-system resilience collaboration, including: Under extreme weather warning, increase the SOC of the energy storage to 95%, fill the thermal storage tank, and start the gas standby unit to run at no load; After the fault, restore the multi-energy supply, preferentially enable photovoltaic-energy storage-heat storage tri-generation units to support key areas, and then gradually restore gas-electricity conversion equipment; Establish a post-disaster evaluation matrix to record power restoration rate, carbon emission reduction loss, and equipment damage index, and optimize the response strategy for the next disaster.

9. A power distribution network dispatch optimization apparatus applying the method of any one of claims 1 to 8, characterized by, It includes: Multi-energy flow data acquisition and synchronization unit, which acquires real-time multi-energy flow operation parameters of the power distribution network to be dispatched, synchronizes data based on a unified timestamp, and cleans up abnormal values to generate a spatiotemporally correlated multi-energy flow dataset, wherein the multi-energy flow operation parameters include node voltage, branch current, photovoltaic output, heat storage tank temperature, heat network flow, and gas pipeline network pressure; The digital twin modeling and dynamic correction unit constructs a dynamic digital twin of the power distribution network to be dispatched, compares the multi-energy flow data set of the power distribution network to be dispatched with the twin simulation results of the dynamic digital twin, and reversely corrects the line impedance parameters and device efficiency coefficients based on device parameter correction rules when the voltage deviation is greater than 2% or the power flow direction is inconsistent, wherein the dynamic digital twin includes a power grid topology, a heat network pipeline transmission delay model, and a gas network node pressure balance equation. The multi-strategy pre-play optimization unit injects preset load fluctuation scenarios and new energy output fluctuation scenarios into the dynamic digital twin, pre-plays multiple dispatch strategies in parallel, and selects the optimal dispatch strategy through weighted scoring. The multi-energy flow collaborative dispatch instruction generation unit generates a multi-energy flow collaborative dispatch instruction set according to the optimal dispatch strategy, wherein the multi-energy flow collaborative dispatch instruction set includes power instructions, heat instructions, and gas instructions. The dynamic task allocation unit dynamically allocates control tasks using edge nodes. The load classification response unit executes differentiated load response in combination with user electricity consumption behavior classification.

10. The power distribution grid dispatch optimization apparatus of claim 9, wherein, Further comprising: The fault detection and emergency recovery unit responds to power distribution network fault events, performs fault detection and emergency recovery, including: Establishing a line temperature dynamic model: T = T amb + I 2 R th ; Wherein, T is the real-time temperature of the line conductor, T amb is the ambient temperature, I is the real-time current, R th is the thermal resistance coefficient; In response to a power distribution network fault event, locate the fault section within 50 milliseconds based on the current abruptness and voltage collapse, preferentially start the island micro-grid to ensure power supply to key areas, and determine whether the line conductor real-time temperature T is greater than 90°C, if so and lasts for 100 milliseconds, then determine that there is a fire risk, immediately disconnect the upstream and downstream breakers at the fault point, activate the dry powder fire extinguishing device, and recheck the fire situation through optical fiber temperature measurement; The closed-loop feedback and model evolution unit feeds back the load response execution amount, fault handling records, and actual carbon emission intensity to the dynamic digital twin, updates the device aging model and iteratively optimizes the dispatch strategy using a model parameter evolution algorithm, and forms a closed-loop autonomous power distribution network operation control system, wherein the model parameter evolution algorithm includes: The principle formula for updating the device aging factor is: η new = η old × [1-β×(t / τ life ) 2 ]; wherein, η old is the initial efficiency of the device, η new is the updated efficiency of the device at the current time, the updated efficiency of the device at the current time is the aging acceleration coefficient, t is the cumulative running time of the device, τ life is the design life of the device; Load elasticity coefficient calibration: when the actual reduction amount and target deviation are greater than 10%, correct according to the following formula; α new = α old × (P real / P target ) wherein, α old is the original load elasticity coefficient, α new is the corrected load elasticity coefficient, P real is the actual load reduction of the user, P target is the target reduction required by the dispatching system; Every 24 hours, perform full model verification on the dynamic digital twin: inject historical fault scenarios to verify the success rate of reconstruction scheme execution; The cross-system resilience collaboration unit includes: Under extreme weather warning, increase the SOC of the energy storage to 95%, fill the heat storage tank of the heat network, and start the gas standby unit for idle operation; After the fault, restore the multi-energy combined supply, preferentially use photovoltaic-energy storage-heat storage three-in-one units to support key areas, and then gradually restore gas-electricity conversion equipment; Establish a post-disaster evaluation matrix to record the power supply recovery rate, carbon emission reduction loss, and device damage index, and optimize the next disaster response strategy.

Images