A method for preventing and controlling fire risks in complex electrical equipment

Abstract

The present invention discloses a method for preventing and controlling fire risks of complex electrical equipment, comprising the following steps: (1) combining historical electrical fire accidents with the working principle of the physical structure of the equipment to construct a risk-side topological network model of the complex electrical equipment system, forming the coupling relationship between risk points and the basic attribute distribution of risk points; (2) calculating the risk propagation Score value between risk points based on the generalized risk fuzzy sorting theory, and quantifying the coupling strength between risks; (3) constructing a complex electrical fire prevention key path generation algorithm that considers the risk propagation characteristics, forming a key propagation path set, thereby achieving prevention and control of complex electrical equipment fires. This method can clarify the accident link of electrical fires in complex electrical equipment and achieve accurate prevention and grasp of the system risk situation. Experimental results show that this method has a good prevention effect and can effectively block fires in complex electrical equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of safety assessment and risk prevention and control of complex electrical equipment systems, and in particular to a method for preventing and controlling fire risks of complex electrical equipment. Background Art

[0002] As a primary energy source in modern industry, agriculture, and social production, electricity has permeated every aspect of the national economy and social life. It is closely intertwined with socioeconomic development and people's lives, becoming essential for social development. However, with the rapid development and increasing modernization of society, the widespread use of various electrical equipment and household appliances has led to a sharp increase in electricity demand. Modern buildings are equipped with a large number of large-scale electrical equipment, which are numerous and widely distributed. Wiring is often concealed within walls, where finishing materials often contain flammable substances. Furthermore, electrical construction processes can be plagued by issues such as improper equipment installation, inappropriate wiring, substandard protective measures, inconsistent product quality, and illegal operation. Furthermore, some electrical wiring, due to age and disrepair, is unable to meet the current peak power demand. This can easily lead to electrical failures and leakage, ultimately resulting in electrical fires. These issues pose a serious threat to public safety and cause significant losses to the national economy.

[0003] Therefore, there is an urgent need to accurately identify and predict the risk points of complex electrical equipment components, clarify the system failure mechanism, determine the key propagation time path after the risk occurs, and realize hierarchical management of risks. Summary of the Invention

[0004] The present invention aims to provide a fire prevention and control method for complex electrical equipment to solve the above problems.

[0005] The technical solution of the present invention is: a method for preventing and controlling fire risks of complex electrical equipment, comprising the following steps:

[0006] Step 1: Based on the historical fault propagation data of complex electrical equipment systems, a risk-side topological network model of the complex electrical equipment system is constructed based on the working principle and causal relationship;

[0007] Step 2: Based on the established network model of the complex electrical equipment system, calculate the coupling strength between risk points, and calculate the risk propagation measure Score value based on the coupling strength of risk points and fuzzy sorting theory;

[0008] Step 3: Study the method of determining the critical path of risk prediction to accurately prevent and control the fire risks of complex electrical equipment systems.

[0009] Preferably, step 1 constructs a risk-side topological network model based on working principles and causal relationships according to the basic risk points, equipment risk points, environmental risk points, management risk points, personnel risk points, and accidents of the complex electrical equipment system, and divides the network model into four risk occurrence layers. The specific steps are as follows:

[0010] The initial causes of electrical fire accident data were mined, and the risk propagation process was deeply analyzed based on causal relationships. The complex electrical equipment system was divided into four risk occurrence layers: personnel layer, facility and equipment layer, environment layer, and management layer. The risk factors of each risk occurrence layer were further refined, specifically as follows:

[0011] At the personnel level, the categories are divided into maintenance personnel, operating staff, and external personnel, and the risk description of the negligent behavior that leads to fire is given for each type of personnel;

[0012] For the facility and equipment layer, it is divided into three subsystems: main equipment, electromechanical, and general equipment. Next, the different subsystems are divided into local components. Then, according to the different components of each local component, it is further divided into various sub-components. Based on the analysis of the working principle of the equipment, the risk description of the accident process caused by the risk hazards of each sub-component is carried out;

[0013] The environmental layer is divided into four main environmental risk factor layers: machine room ambient temperature, structural water leakage, condensation water, and brake shoe dust. The risk of accidents caused by each risk factor is described.

[0014] For management, the causes of accidents are divided into four aspects: rules and regulations, maintenance management, supervision and inspection, and risk control, and a risk description is given for each cause of the accident.

[0015] Preferably, step 2 adopts a parameter characterization method for the degree of influence between risk points based on fuzzy sorting to perform fuzzy risk quantitative sorting, calculate the risk propagation measurement Score value from the upstream risk point to the adjacent downstream risk point, and obtain the degree of influence value of the associated risk point, which specifically includes the following steps:

[0016] Step 2.1: Calculate the normalized generalized fuzzy number of upstream risk points

[0017]

[0018] Among them: a i1 Indicates the first score of the fuzzy evaluation of the upstream risk point, |a ij |is a ij The absolute value of Indicates a ijRound up, 1≤i≤n, j=1,2,34, Indicates the pre-set weights, represents the normalized generalized fuzzy number The corresponding element in

[0019] Step 2.2: Calculate the normalized generalized fuzzy number of upstream risk points Standard deviation

[0020]

[0021] in, represents the normalized generalized fuzzy number The corresponding element in

[0022] Step 2.3: Calculate the normalized generalized fuzzy number of upstream risk points The left side area of ​​the trapezoid Area of ​​the left side of the right waist of the trapezoid Area of ​​the right side of the left waist of the trapezoid and the right side area of ​​the trapezoid

[0023]

[0024]

[0025] Step 2.4: Calculate the normalized generalized fuzzy number of upstream risk points Risk Communication value;

[0026]

[0027] in,

[0028] Step 2.5: Add the Score value to the score of the downstream risk point to obtain the risk level of the downstream risk point.

[0029] Preferably, step three proposes a complex electrical fire prevention critical path generation algorithm based on Dijkstra and risk propagation, transforming the risk propagation maximum probability calculation problem into a shortest path solution problem and evaluating key nodes, specifically:

[0030] Step 3.1: Set the risk from the initial risk point S to the end risk point v i The path of ∈V can only pass through the terminal risk points that are already included in the set S, and cannot pass through the remaining terminal risk points that are not yet in the set S;

[0031] Step 3.2: dist[s, vi ] represents the length of the shortest path relative to the set S, short[s]=short[s,v i ] represents the global risk from the initial risk point to the end risk point v i ∈V\{S}, which stipulates that this shortest path has no restriction that "it must only pass through the terminal risk points that are included in the set S in the middle of the path". The global shortest paths formed at this time are:

[0032] dist[s,v i ]≥short[s,v i ]

[0033] Step 3.3: Continuously calculate dist[s, v] through Dijkstra algorithm i ] Then the set S is continuously expanded. When the set S is continuously expanded, the shortest path relative to the set S will become shorter and shorter until v i When entering the set S, there is dist[s,v i ]=short[s,v i ];

[0034] Step 3.4: Decompose the entire risk network propagation model into independent sub-components. Within each sub-component, manually control the specific status of a basic attribute point to re-evaluate its probability of occurrence. For each changed basic attribute point within the sub-component, re-apply fuzzy interval theory to calculate the risk propagation probability value of each edge on each risk propagation path involved. Re-apply the key risk path determination algorithm to derive a new risk propagation path for the sub-component. After changing the experimental parameters of several groups of basic attribute points, new key risk propagation paths are re-derived, and all key risk propagation paths of the four sub-components are aggregated into the entire risk propagation network model.

[0035] The beneficial effects of the present invention are:

[0036] The present invention establishes a risk relationship network model based on a detailed analysis of the specific causes and risk data of typical complex electrical fires, and studies the risk assessment, risk prevention, and post-risk treatment measures of electrical fires. It achieves accurate identification and prediction of the risk situation of complex electrical equipment systems, clarifies the system failure mechanism, and the failure path that can prevent the spread of risks after they occur, and realizes risk classification management, hidden danger investigation, and active safety prevention and control of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A flowchart of a method for preventing and controlling fires in complex electrical equipment provided by an embodiment of the present invention;

[0038] Figure 2 A risk propagation process diagram of a complex electrical equipment fire prevention and control method provided by an embodiment of the present invention;

[0039] Figure 3 A simplified diagram of the propagation path of a method for preventing and controlling fires in complex electrical equipment provided by an embodiment of the present invention;

[0040] Figure 4 This is a comparison chart before and after the experiment of each sub-part of a complex electrical equipment fire prevention and control method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The present invention will be further described below with reference to the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it. The embodiments of the present invention are not limited thereto.

[0042] Example 1

[0043] like Figure 1 As shown, a method for preventing and controlling fires in complex electrical equipment includes the following steps:

[0044] Step 1: Construct a risk transfer network model based on the basic risk points of complex electrical equipment systems, equipment risk points, environmental risk points, management level risk points, personnel level risk points and accidents;

[0045] We collected data on electrical fire accidents from 2005 to 2019, analyzed the initial causes of each fire accident, and identified the specific process of risk propagation. We identified four primary risk levels within complex electrical equipment systems: personnel, facilities, equipment, environment, and management. We then further categorized each risk level.

[0046] At the personnel level, the categories are divided into maintenance personnel, operating personnel, and external personnel. Risk descriptions are provided for each type of personnel's negligent behavior that may lead to fire.

[0047] At the facility and equipment level, it is first divided into three subsystems: main equipment, electromechanical, and general equipment. Next, the different subsystems are divided into local components. Then, according to the different components of each local part, it is further divided into various sub-components, and a risk description is made for the accident process caused by the risk hazards of each sub-component.

[0048] For the environmental level, it is divided into four main environmental risk factors: machine room ambient temperature, structural water leakage, condensation water and brake shoe dust, and a risk description of the accidents caused by each risk factor is given.

[0049] At the management level, the causes of accidents are divided into four aspects: rules and regulations, maintenance management, supervision and inspection, and risk control. The risk of accidents caused by each cause is described.

[0050] Step 2: Based on the established network model of the complex electrical equipment system, the coupling strength between risk points is calculated. Based on the coupling strength of risk points and fuzzy sorting theory, a fuzzy risk quantification sorting model is proposed to calculate the risk propagation measure Score value;

[0051] The main parameters and calculation methods of the fuzzy risk quantification ranking model are as follows:

[0052] set up is a generalized fuzzy number, a i ∈R,b i ∈R,

[0053] The definition is as follows:

[0054] Generalized risk fuzzy number addition

[0055]

[0056] Generalized risk fuzzy number multiplication method

[0057]

[0058] Generalized risk fuzzy number division method

[0059]

[0060] Among them: b1≠0, b2≠0, b3≠0, b4≠0.

[0061] Calculate the standard deviation of the trapezoidal fuzzy model

[0062]

[0063] in:

[0064] Converting generalized risk fuzzy numbers into standardized generalized fuzzy numbers

[0065]

[0066] in: |a ij |is a ij The absolute value of Indicates a ijRound up, 1≤i≤n, j=1, 2, 3, 4.

[0067] Calculate the area on the left side of the trapezoid of the standardized generalized risk fuzzy number

[0068]

[0069] The area on the left side of the right waist of the trapezoid of the standardized generalized risk fuzzy number

[0070]

[0071] The area on the left side and right side of the trapezoid of the standardized generalized risk fuzzy number

[0072]

[0073] The right side area of ​​the trapezoid of the standardized generalized risk fuzzy number area:

[0074]

[0075] The risk propagation measure Score value of the standardized generalized risk fuzzy number is:

[0076]

[0077]

[0078] Step 3: Determine the key paths of risk propagation to achieve precise prevention and control of complex electrical fires.

[0079] Define a set S, if a termination risk point v in the set V\{S} i In the set S, from the starting risk point to the ending risk point v i The shortest path ∈V\{S} has been found, and at the beginning of the algorithm, there are only starting risk points in the set S. That is:

[0080] S:={v i ∈V: the shortest path of vertex v i has been found}

[0081] The algorithm is completed if and only if S = V. At this point, all elements in set V are put into set S, which means that the shortest paths from the starting risk point to all other ending risk points except the starting risk point have been found.

[0082] From the starting risk point to the ending risk point v i∈V(at the beginning ) relative to the shortest path of the set S. That is, from the starting risk point to the ending risk point v i The path of ∈V can only pass through the terminal risk points that are already included in the set S, and cannot pass through the other terminal risk points that are not yet in the set S. The length of the shortest path relative to the set S is recorded as: dist[s,v i ], and the previous short[s]=short[s,v i ] represents the global risk from the starting risk point to the ending risk point v i ∈V\{S}, this shortest path does not have the restriction that "it must only pass through the terminal risk points that are already included in the set S". This global shortest path is the final solution. Therefore, there is generally a relationship:

[0083] dist[s,v i ]≥short[s,v i ]

[0084] What Dijkstra algorithm has to do is to calculate dist[s,v i ] Then the set S is continuously expanded. When the set S is continuously expanded, the shortest path relative to the set S will become shorter and shorter until v i When entering the set S, we get short[s,v i ], and at this time dist[s,v i ]=short[s,v i ].

[0085] The entire risk network propagation model is decomposed into four independent sub-parts. Within each sub-part, the specific status of a basic attribute point is manually controlled to re-evaluate its probability of occurrence. Fuzzy interval theory is re-applied to each changed basic attribute point in the sub-part to calculate the risk propagation probability value of each edge on each risk propagation path involved. The key risk path determination algorithm is re-applied to derive a new risk propagation path for the sub-part. After changing the experimental parameters of several groups of basic attribute points, a new key risk propagation path is re-derived, and all key risk propagation paths of the four sub-parts are aggregated into the entire risk propagation network model.

[0086] We found that in the four sub-sections of the risk propagation network model, manually controlling the initial states of related basic attribute points or other controllable risk points led to a decrease in the overall risk probability of the risk propagation path where the controlled risk points were located. This demonstrates that human control does play a role in preventing risks from occurring.

[0087] Example 2

[0088] like Figure 2 Figure 1 shows an example of a risk propagation process. By collecting data on complex electrical equipment, such as electrical fires on urban rail trains, all accident data was extracted and separated to obtain risk propagation paths. All risk propagation paths were then comprehensively organized to form a risk propagation network model.

[0089] On the basis of the risk propagation network model, a detailed classification is carried out on all risk points in the network model. According to their characteristics, they are divided into basic attribute points, state points with divisible attributes, physical points with divisible attributes, physical points with indivisible attributes, state points with indivisible attributes, environmental factor points, management level risk points, and personnel level risk points. And they are marked on the network risk model according to different legend colors. Basic attribute points come from state points or physical points with divisible attributes. In the present invention, there are 12 types of points for attribute point division. In the entire network model, each point represents the source of the risk hazard, or represents a certain state of the risk source point in the process of transitioning from a certain risk source point to the final accident point. Each edge in the network model represents the risk propagation process, and the propagation direction is the direction of the arrow.

[0090] Taking a risk transmission path as an example, before calculating the fuzzy number, it is necessary to map the linguistic terms to the corresponding fuzzy intervals. Tables 1 and 2 show the fuzzy numbers corresponding to the linguistic terms, and Tables 3 to 7 show the calculation process from the first to the fifth level respectively.

[0091] Table 1 Risk likelihood and risk consequences Linguistic terms and their corresponding fuzzy numbers (i=A,B, )

[0092]

[0093] Table 2 Risk levels Linguistic terms and their corresponding fuzzy numbers

[0094]

[0095] Table 3 First layer calculation process

[0096]

[0097] Step 1: According to formulas (1)-(3), calculate a risk point generalized fuzzy number:

[0098]

[0099] Based on formula (5), the generalized fuzzy number Convert to normalized generalized fuzzy number

[0100]

[0101] Step 2: According to formula (4), calculate a normalized generalized fuzzy number Standard deviation

[0102]

[0103] Step 3: Based on formulas (6)-(9), calculate a normalized generalized fuzzy number Area and

[0104]

[0105] Step 4: Based on formulas (11) and (12), calculate a normalized generalized fuzzy number of and

[0106]

[0107] Step 5: According to formula (10), calculate a standardized generalized fuzzy number of The larger the value of , the higher the risk probability of the risk point.

[0108]

[0109] After processing the Score value, the risk level of the next risk point C50 and C49 is obtained 0.732170488649139

[0110] Table 4 Second layer calculation process

[0111]

[0112] The same goes for

[0113] (1) and C54 risk level 0.512863683556364

[0114] (2) The Score value is the same After processing, the next risk point B46 risk level is obtained 0.688995635884659

[0115] Table 5 The third layer calculation process

[0116]

[0117] Similarly:

[0118] (1)

[0119] (2)

[0120] (3)

[0121] Table 6 The calculation process of the fourth layer

[0122]

[0123] get After processing the Score value, the risk level of the next risk point B69 is obtained:

[0124]

[0125] Table 7 Fifth layer calculation process

[0126]

[0127] like Figure 3 As shown, this is one of the subgraphs. The subgraphs are numbered and v1 is selected as the origin S. Then at the beginning of the algorithm, S = {v1}.

[0128] At this time there are:

[0129] dist[v1, v2] = ω 1,2 = 0.5, dist[v1, v3] = 0.5

[0130] dist[v1,v4]=dist[v1,v5]…=dist[v1,v 11 ]=∞

[0131] v2, v3 have the highest probability of being in v1, so put v2 and v3 into set S:

[0132] S = {v1, v2, v3}

[0133] Continue the while loop, now we have:

[0134] dist[v1, v2] = ω 1，2 = 0.5, dist[v1, v3] = 0.5

[0135] dist[v1, v4] = ω 1，2 *ω 2，4=0.116085

[0136] dist[v1, v5] = ω 1，2 *ω 2,5 =0.116085

[0137] dist[v1, v9] = ω 1,3 *ω 3,9 =0.014425

[0138] dist[v1,v6]=dist[v1,v7]=…=dist[v1,v 11 ]=∞

[0139] v4 and v5 have the highest probability of being in v1, so put v4 and v5 into set S:

[0140] S = {v1, v2, v3, v4, v5}

[0141] Continue the while loop until you have:

[0142] dist[v1, v2] = ω 1，2 = 0.5, dist[v1, v3] = 0.5

[0143] dist[v1, v4] = ω 1，2 *ω 2，4 =0.116085

[0144] dist[v1, v5] = ω 1,2 *ω 2,5 =0.116085

[0145] dist[v1, v9] = ω 1，3 *ω 3，9 =0.014425

[0146] dist[v1,v6]=ω 1，2 *ω 2，4 *ω 4，6 =0.038568

[0147] dist[v1, v7] = ω 1，2 *ω 2，5 *ω 5，7 =0.001118

[0148] dist[v1, v8] = ω 1，2 *ω 2，5 *ω 5，8 =0.001118

[0149] dist[v1,v 10 ]=ω1，2 *ω 2，5 *ω 5,10 =0.001118

[0150] dist[v1,v 11 ]=ω 1，3 *ω 3，9 *ω 9,11 =0.000026

[0151] dist[v1,v 11 ]=ω 1，2 *ω 2,5 *ω 5,7 , *ω 7,11 =0.000006

[0152] dist[v1,v 11 ]=ω 1，2 *ω 2，5 *ω 5,8 *ω 8,11 =0.000088

[0153] dist[v1,v 11 ]=ω 1，2 *ω 2，5 *ω 5,10 *ω 10，11 =0.000088

[0154] Because dist[v1,v 11 ]=ω 1，2 *ω 2，5 *ω 5，7 *ω 7，11 =0.000006<0.00048;

[0155] dist[v1,v 11 ]=ω 1,2 *ω 2,5 *ω 5,8 *ω 8,11 =0.000088<0.00048;

[0156] dist[v1,v 11 ]=ω 1,2 *ω 2,5 *ω 5,10 *ω 10,11 =0.000088<0.00048,

[0157] Therefore, no update

[0158] v 11 The probability of v1 is the largest, so v 11 Put into set S:

[0159] S={v1,v2,v3,v4,v5,v6,v9,v7,v8,v 10 ,v 11}

[0160] This is the last time, and no updates have occurred, and at this time:

[0161]

[0162] The condition of the while loop in the algorithm is not satisfied, the loop ends, and the algorithm ends. Obviously, V = S at this time.

[0163] So maxdist[v1,v 11 ]=0.00048, that is, v1→v 11 The maximum probability path is: v1→v2→v4→v6→v 11 By changing the parameters of some controllable basic attribute points, the overall impact of these factors on the central risk point can be altered. Consequently, the probability values ​​along certain risk transmission paths will change, and so will the key risk transmission paths. For each set of data, the changes in basic attribute points serve as independent variables, and the changes in probability values ​​and key risk transmission paths serve as dependent variables. This yields multiple sets of experimental results. By analyzing the experimental results in line graphs, we can determine the varying effects of artificially altering basic factors on the entire network model, ultimately establishing the relationship between the independent and dependent variables and drawing relevant conclusions.

[0164] During the experimental design process, the entire risk network propagation model was decomposed into four independent sub-parts. Within each sub-part, the specific status of a basic attribute point was manually controlled to re-evaluate its probability of occurrence. Fuzzy interval theory was re-applied to each changed basic attribute point in the sub-part to calculate the risk propagation probability value of each edge on each risk propagation path involved. The key risk path determination algorithm was re-applied to derive the new risk propagation path for the sub-part. After changing the experimental parameters of several groups of basic attribute points, the new key risk propagation path was re-derived, and all the key risk propagation paths of the four sub-parts were summarized into the entire risk propagation network model.

[0165] like Figure 4 As shown, the line graph comparing the probability of occurrence of risk points on the edge of the risk propagation path before and after each part of the experiment: the entire experiment is also divided into four parts according to the division method of the overall risk network propagation model. For each specific experiment in each group of experiments, the initial changes in the basic attribute points and the specific probability value changes caused by them on the corresponding risk propagation path will be displayed.

[0166] By analyzing and comparing the line graphs of the different experimental results above, we found that in the four sub-components of the risk propagation network model, manually controlling the initial states of related basic attribute points or other controllable risk points led to a decrease in the overall risk probability of the risk propagation path where the controlled risk points were located. This demonstrates that human control does play a role in preventing risks from occurring.

[0167] By comparing the specific number and distribution of key risk paths within each sub-section and across the entire risk spectrum before and after the experiment, we found that only some of the nodes on each risk path changed, with the overall change being insignificant. Furthermore, while human control can reduce the overall probability of risk occurrence on each risk transmission path, the number and overall distribution of key risk transmission paths remain essentially unchanged. Therefore, strict control can be implemented on certain essential points along certain risk transmission paths to effectively prevent risks.

[0168] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of an embodiment, and the processes in the accompanying drawings are not necessarily required to implement the present invention.

Claims

1. A method for preventing and controlling fire risks of complex electrical equipment, characterized in that: The steps include: Step 1: Based on the historical fault propagation data of complex electrical equipment systems, a risk-side topological network model of the complex electrical equipment system is constructed based on the working principle and causal relationship; Step 2: Based on the established network model of the complex electrical equipment system, calculate the coupling strength between risk points, and calculate the risk propagation measure Score value based on the coupling strength of risk points and fuzzy sorting theory; Step 3: Research methods for determining critical paths for risk prediction to accurately prevent and control fire risks in complex electrical equipment systems; Among them, step 2 adopts the parameter characterization method of the degree of influence between risk points based on fuzzy sorting to perform fuzzy risk quantitative sorting, calculate the risk propagation measurement Score value from the upstream risk point to the adjacent downstream risk point, and obtain the degree of influence value of the associated risk point. Specifically, it includes the following steps: Step 2.1: Calculate the normalized generalized fuzzy number of upstream risk points Among them: a i1 Indicates the first score of the fuzzy evaluation of the upstream risk point, |a ij |is a ij The absolute value of Indicates a ij Round up, 1≤i≤n, j=1,2,3,4, Indicates the pre-set weights, represents the normalized generalized fuzzy number The corresponding element in Step 2.2: Calculate the normalized generalized fuzzy number of upstream risk points Standard deviation in, represents the normalized generalized fuzzy number The corresponding element in Step 2.3: Calculate the normalized generalized fuzzy number of upstream risk points The left side area of ​​the trapezoid Area of ​​the left side of the right waist of the trapezoid Area of ​​the right side of the left waist of the trapezoid and the right side area of ​​the trapezoid Step 2.4: Calculate the normalized generalized fuzzy number of upstream risk points Risk Communication value; in, Step 2.5: Add the Score value to the score of the downstream risk point to obtain the risk level of the downstream risk point.

2. A method for preventing and controlling fire risks of complex electrical equipment according to claim 1, characterized in that: Step 1: Based on the basic risk points, equipment risk points, environmental risk points, management level risk points, personnel level risk points, and accidents of complex electrical equipment systems, a risk-side topological network model based on working principles and causal relationships is constructed, and the network model is divided into four risk occurrence layers. The specific steps are as follows: The initial causes of electrical fire accident data were mined, and the risk propagation process was deeply analyzed based on causal relationships. The complex electrical equipment system was divided into four risk occurrence layers: personnel layer, facility and equipment layer, environment layer, and management layer. The risk factors of each risk occurrence layer were further refined, specifically as follows: At the personnel level, the categories are divided into maintenance personnel, operating staff, and external personnel, and the risk description of the negligent behavior that leads to fire is given for each type of personnel; For the facility and equipment layer, it is divided into three subsystems: main equipment, electromechanical, and general equipment. Next, the different subsystems are divided into local components. Then, according to the different components of each local component, it is further divided into various sub-components. Based on the analysis of the working principle of the equipment, the risk description of the accident process caused by the risk hazards of each sub-component is carried out; The environmental layer is divided into four main environmental risk factor layers: machine room ambient temperature, structural water leakage, condensation water, and brake shoe dust. The risk of accidents caused by each risk factor is described. For management, the causes of accidents are divided into four aspects: rules and regulations, maintenance management, supervision and inspection, and risk control, and a risk description is given for each cause of the accident.

3. A method for preventing and controlling fire risks of complex electrical equipment according to claim 1, characterized in that: Step 3 proposes a critical path generation algorithm for complex electrical fire prevention based on Dijkstra and risk propagation, transforming the maximum probability calculation problem of risk propagation into a shortest path problem and evaluating key nodes. Specifically: Step 3.1: Set the risk from the initial risk point S to the end risk point v i The path of ∈V can only pass through the terminal risk points that are already included in the set S, and cannot pass through the remaining terminal risk points that are not yet in the set S; Step 3.2: dist[s,v i ] represents the length of the shortest path relative to the set S, short[s]=short[s,v i ] represents the global risk from the initial risk point to the end risk point v i ∈V\{S}, which stipulates that this shortest path has no restriction "must pass through only the terminal risk points included in the set S in the middle of the path". The global shortest paths formed at this time are: dist[s,v i ]≥short[s,v i ] Step 3.3: Continuously calculate dist[s,v i ] Then the set S is continuously expanded. When the set S is continuously expanded, the shortest path relative to the set S will become shorter and shorter until v i When entering the set S, there is dist[s,v i ]=short[s,v i ]; Step 3.4: Decompose the entire risk network propagation model into independent sub-components. Within each sub-component, manually control the specific status of a basic attribute point to re-evaluate its probability of occurrence. For each changed basic attribute point within the sub-component, re-apply fuzzy interval theory to calculate the risk propagation probability value of each edge on each risk propagation path involved. Re-apply the critical risk path determination algorithm to derive the new risk propagation path for the sub-component. After changing the experimental parameters of several groups of basic attribute points, new key risk propagation paths are derived, and all key risk propagation paths of the four-character parts are summarized into the entire risk propagation network model.

Images