JPH09231282A - Method for emergency network operation - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for emergency network operation which flexibly and speedily forms operation plans in response to various emergency states. SOLUTION: To plan operation in case of emergency different from ordinary time in a transportation network wherein plural supply points and consumption points are represented as nodes and paths for transporting goods between nodes are represented as arcs, an instance data base 6 stored with instances consisting of operation methods and instance indexes in case of emergency which are obtained by previous examination and a cost data base 30 stored with the costs for linking with the operation methods of the instances are provided. In case of emergency, a current feature quantity is found by analyzing the state at the emergency time (2), retrieval from the instance data base is performed according to the current state feature quantity (4), an applied instance is determined (8); and the cost for linking with the operation method of the extracted instance is extracted from the cost data base (18 and 20), and various data needed for the extracted cost and planning are inputted to form an operation plan by distribution programming (16).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention makes an optimum operation plan when an abnormal situation occurs in a transportation route, a resource supply point, etc. in a water supply pipe network, a gas supply network, an article distribution network, etc. Emergency network operation method.

[0002]

2. Description of the Related Art Transportation network systems that have many points of supply and consumption of resources are used in various fields of modern society regardless of fields, such as water supply pipe networks, gas and power supply networks, and material distribution networks. Used in industry. Generally, such a network system is large-scale and complicated, and it is a very difficult problem to operate and manage the whole system efficiently. In particular, public facilities such as water and electricity are indispensable for people to live as lifelines, so they have a duty to stably supply resources under any circumstances. ing.

The problem of planning the supply amount and the transportation amount according to the demand is to set the transportation cost for each arc in the transportation network in which the supply point and the relay point are the nodes and the arcs connect the nodes. It can be solved as a minimum cost flow problem that calculates the transportation amount that minimizes the total cost.
Furthermore, a spatial network is layered over time to form a multi-layered extended network model, and an arc is also extended from the storage point to the storage point of the next layer, and the storage amount at each time is determined by the transport amount of the arc that extends between layers. By expressing it, the dynamic allocation problem can be reduced to the minimum cost flow problem and solved. In Japanese Patent Application Laid-Open No. 5-35714 “Network Operation Method”, by using this multi-layered extended network, not only the constraint conditions such as upper and lower limit values are satisfied, but also a discrete transportation amount that reduces the time fluctuation of the transportation amount is also used. It describes a method to solve at high speed while satisfying realistic requirements such as handling.

As for network operation control including emergency, Hasumoto, Yamagishi, Mori: "Application of water operation integrated management system" 4th environmental system automatic measurement control domestic workshop papers pp. 154-157 (1992.9),
It describes the water operation simulation to support the calculation of the downtime of the water supply facilities and the method to digitize the expert knowledge of the operator as a knowledge base and apply the artificial intelligence technology to provide the emergency guidance.

[0005]

Most of the conventional large-scale transportation network operation management methods are intended only for normal operation management. Even in the distribution plan that seeks the minimum cost flow using the above-mentioned multi-layered extended network, if the system is constructed assuming normal times, only one type of cost needs to be prepared, but this is operated in an emergency. Even if it is used for planning, it does not always derive an appropriate solution. Even if multiple costs are prepared, which one will be applied becomes a problem, and quick planning cannot be achieved. In addition, when the operation know-how in an emergency is heard from the operator and the rule group is stored as a knowledge base and an operation plan is drafted using this, it takes a very long time to construct the knowledge base. However, it is extremely difficult to express ambiguous know-how explicitly. Furthermore, it is impossible to infer the matters that are not in the knowledge base, and at present, it is not possible to support the planning so as to satisfy the operator's request. An object of the present invention is to provide an emergency network operation method that flexibly and promptly prepares an operation plan against occurrence of various emergency situations.

[0006]

In order to achieve the above object, the present invention is normally used in a transportation network in which a plurality of supply points and consumption points are represented as nodes and a route for transporting a substance between the nodes is represented as an arc. It is an emergency network operation method for planning an operation plan in an emergency when a situation different from the time occurs. A case database that stores cases that form a case index is searched by the current feature amount, the case corresponding to the case index that is most similar to the current feature amount is extracted, and the cost linked to the operation method of the case is stored. The cost linked to the operating method of the extracted case is extracted from the cost database, and the extracted cost and various data necessary for planning are extracted. As an input, the operation plan is prepared in the allocation plan. Further, in the distribution plan, a cost is allocated to each of the arcs to obtain a minimum cost flow, and a flow rate correction process is performed according to setting of a target value that satisfies a practical operational requirement item, and the process result is When the required items are not satisfied in operation, the flow rate correction process is performed again according to the setting change of the target value. In the preliminary examination, the allocation plan is executed according to the setting of operation request items and the setting of costs, and when the execution result is satisfactory, the case where the operation request items are set and the set cost are respectively set. The allocation plan is stored in the case database and the cost database, and when the execution result is not satisfactory, the cost setting or the operation request item setting is changed to execute the distribution plan. When a case in the case database is specified, the case index and operation method of the case and the cost in the cost database linked to the operation method are displayed. In addition, the present state feature amount is obtained by analyzing the present state feature amount in an emergency, and a case database that stores a case that is a set of an operation method and a case index in an emergency obtained by preliminary examination is searched by the present state feature amount and the present state feature amount is searched. The case corresponding to the case index that is most similar to the feature quantity is extracted and linked to the extracted case operation method from the rule database that stores the general operation rules generated based on the case operation method and background knowledge. The operation rules to be used are extracted, and the extracted operation rules are used as an operation plan for network operation. In addition, a current state feature amount is obtained by an analysis of the current state of the emergency, and a case database that stores a case that is a set of an emergency operation method and a case index obtained by preliminary examination is searched by the current state feature amount, The case corresponding to the case index that is most similar to the feature amount is extracted, the operation method of the extracted case and the current data are fused to create an initial operation plan, and the created initial operation plan is evaluated by simulation and evaluated. If the result is not satisfied, the created initial operation plan is modified, or the cost linked to the extracted operation method of the case is extracted from the cost database storing the cost linked to the operation method of the case, and the extraction is performed. Executing the allocation plan based on the cost that was created and creating a plan, and simulating the modified initial operation plan or the planned plan again. Assessed by ® down,
If the result of the evaluation shows that it is satisfactory, the modified initial operation plan or the prepared operation plan is used as the operation plan for network operation.

[0007]

Embodiments of the present invention will be described with reference to the drawings. FIG. 1 shows a processing flow of emergency operation planning according to the present invention. First, analyze the current situation to see what kind of emergency is occurring (step 2),
Understand the situation. Current situation analyzed is case database 6
It is used to measure the similarity between the case and the current situation when searching for cases stored in it (step 4), and the most similar case is determined as the application case (step 8), and the distribution plan is based on this. Execute 16. In the distribution plan 16, the cost is set as a function of the transportation amount in each of the arcs 68 to 91 in the multi-layer expansion network (for example, in the case of the water supply pipe network, a function in which the amount of water flowing within a certain time is a variable is set). Solve the minimum cost flow problem. The cost is automatically set as a part of the input data (step 20) by calling from the cost database 30 the cost linked to the operation method belonging to the case determined as the current application case (step 18).

In the present embodiment, the cost, which is a part of the input data of the distribution plan 16, is characterized in that the one linked to the case is automatically called. Since the cost is a numerical expression of the problem solving method in general, its usage is not limited, and it has the great advantage that it has the effect of flexibly filling the gap between the current situation and the case. An operation plan suitable for each situation can be quickly and without being unsolvable without trial and error to fill it.

In the emergency situation analysis (step 2), the situation is represented by a situation attribute. The attributes clearly indicate characteristics of an emergency situation such as accident type, accident facility, accident degree, accident occurrence time period, accident consumption, and the like. The case database 6 is searched using the attribute of the situation obtained in the current situation analysis (step 2) (step 4).

FIG. 2 shows details of the case database 6.
The case database 6 normally considers an operation plan by using the distribution plan 16 assuming the accident facility and the accident degree (step 1), and the case index 24 and the operation method 26.
It is a database that collects the case examples 22 that form a set. Here, the case index 24 characteristically expresses the emergency situation assumed in the case 22, and matches the situation attribute used in the present situation analysis 2, or has a deep relationship with it, and the relationship between the two is Must be clear. Further, the normal time is also stored in the case database 6 as one case. In the case search 4, the current attribute and the case index are compared to determine the similarity, and the case to which the most similar case index belongs is determined as the application case (step 8). Similar definitions are likely to differ depending on the application, but as a simple example, the distance can be defined as in the following equation and the one with the smallest distance can be defined as the case that is most similar to the current state. .

[0011]

[Equation 1]

Here, i is an index, n is the number of attributes, and a
i is the current value of the attribute i, xi is the value of the case attribute i, and d is the distance between two points in the feature space spanned by each attribute.

FIG. 9 shows an example of a case where means for displaying the cases in the case database 6 is added to the actual system. As an example, assume a window-type user interface. For example, if the case number is specified by pressing the button of the case content display 21, the case 22 is displayed. This includes a case index 24 and an operation content 26. The present embodiment is characterized in that there is a button 27 for displaying a cost 28 linked to it in the operation content 26, and a cost display display for displaying at least a part of the cost of linking immediately if necessary. You can access 29. The costs are sequentially displayed by scrolling the screen.

FIG. 3 shows a processing flow of the distribution plan 16,
FIG. 4 shows a conceptual diagram of the multilayer extended network. Allocation plan 16 is basically minimum cost flow calculation 3 by mathematical programming.
Achieved by 4. Spatial networks consisting of a plurality of storage points 56 to 58 and nodes 65 and arcs 80 to 83 connecting them are layered in layers for each hour, and the amount of transportation on the arcs 68 to 79 connecting the same storage points between layers is every hour. By assigning a cost that is a function of the transportation amount to all the arcs 68 to 91, the method that is known in the field of mathematical programming as a minimum cost flow problem that seeks a transportation flow that minimizes the total cost value. Can be used to solve dynamic allocation planning problems. Thereby, for example, in the case of a water supply pipe network of water supply, the water supply amount of each water supply pipe is obtained.

The following input data is required to obtain the minimum cost flow (step 34). 1) Network configuration information 2) Initial storage amount at storage point 3) Supply amount 4) Consumption node and consumption amount 5) Upper and lower limit of transportation amount at each arc 6) Cost as a function of transportation amount at each arc 7 ) The minimum cost flow is output by setting data above the facility conditions other than the upper and lower limits required for transportation and storage (step 32) and inputting the data to the minimum cost flow calculation 34 by mathematical programming. However, the plan value at this point is merely the flow in which the total cost value is the minimum among those satisfying various constraints. In the actual problem, it is difficult to rapidly change the water purification capacity and the output of the water supply pump, for example, in the case of the water supply pipe network of water supply, and the burden on the operator who operates the facility in this way also increases. Therefore, it is required in actual operation to impose restrictions on the time fluctuations of transportation volume. Therefore, 8) flow rate correction target transportation route 9) correction priority order 10) time variation allowable range of flow rate 11) other special constraint conditions are set as various conditions required for operation (step 3).
6) Input to the solution improvement processing 38.

FIG. 5 shows a processing flow performed in the solution improving processing 38. This is a step of setting a target value that satisfies various conditions required for operation and correcting the flow rate so that the planned value is as close as possible to the target value. In the present embodiment, after the flow rate correction processing 46 is once performed in this processing, the target value is changed (step 50), and the flow rate correction processing 46 is performed again.
It is characterized in that it prepares a part to do. First, an arbitrary closed loop including the flow rate correction target transportation route is listed in the multilayer extended network (step 42). Next, the target value is automatically set so as to satisfy the various conditions required for operation while maintaining the total flow rate of each flow rate correction target transportation route output by the minimum cost flow calculation (step 44). For example, when the condition required for operation is to keep the flow rate constant, the time average of the total flow rates obtained by the plan becomes the target value. The flow rate is corrected so that the target value thus obtained is achieved (step 46). The difference between the planned value and the target value at that time is used as the correction amount.If the arc direction on the closed loop prepared in advance is the same direction as the loop, a positive correction amount is used. Add a negative correction amount to the planned amount. If the correction process is performed on an arbitrary closed loop including the flow rate correction target transportation route, the significance of obtaining the minimum cost flow is diminished. Therefore, only for a loop in which the amount of change in cost due to the correction is 0 or less than a predetermined value. It is desirable to perform step 46. In addition, facility restrictions are given priority over operational requirements, and corrections are made within the facility restrictions. From the output of the above processing, how much the various conditions required for operation are satisfied is evaluated (step 48), and if satisfied, the solution improving processing ends. If it is not satisfied, it is considered that there is a problem with the target value for which the cause is set, and this is changed (step 50). In fact, if there is a large gap between the facility constraint and the target value in a certain time zone, the location where the flow rate changes abruptly is also output. However, since it is not known whether the target value is truly inappropriate unless the correction processing 46 is actually performed, the flow rate correction processing 46 is performed once, and then the target value is changed (step 50), and the flow rate is again measured. The correction process 46 is performed, and the solution improvement process 38 ends.

The cost used in the distribution plan 16 is examined in advance 1.
It is also necessary to link with the case and determine the cost to be used in the allocation plan 16 at the same time as the application case is determined. In the present embodiment, the distribution plan 1 is used not only in an actual emergency but also in the preliminary examination 1.
The feature is that the cost linked to the case is set by using 6. FIG. 6 illustrates the processing flow in the preliminary examination 1. First, operation request items are selected (step 100). Instead of selecting the operation request item, if there is a specific operation example such as a past case, the operation plan may be selected (step 102). The cost is appropriately set so that the selected operation is realized (step 104). This is input to the distribution plan 16 and the output is evaluated (step 108). If the output satisfies the condition initially selected, the cost set in the cost setting is stored in the cost database in association with the operation plan selected in advance (step 110). If the requirement is not satisfied, steps 104 and 16 are repeated until the cost is changed and satisfactory operation is obtained. If the requirement cannot be satisfied simply by changing the cost, consider the requirement to be inappropriate and revise the target operation.

FIG. 7 shows a display example required for setting the cost linked to the case in the preliminary examination 1. The present embodiment is characterized in that the buttons 121 to 124 are displayed when the preliminary examination execution button 120 is pressed. Button 1
When 21 is pressed, a screen for setting the target operation in the emergency situation of interest is displayed. Here, step 100 or step 102 is processed. When the button 122 is pressed, the cost setting editor 130 is displayed, which can be used to set the cost function. This corresponds to step 104.
When the cost setting is completed, the button 123 is pressed to execute the distribution plan 16. When the desired cost is set, the button 124 is pressed to store the cost (step 110).

As an alternative to the above-mentioned preliminary study 1, explanation-based learning (EBL) 14
We propose a method to generalize the operation method using 0, store it by linking it with the case, and perform a distribution plan using the generalized operation rule linked to the application case when planning. The cost is a numerical expression of the operating method, while the operating rule is a logical expression of the operating method. An example is shown in FIG.
Although it is similar to FIG. 1, it differs in that an operation rule is used instead of cost to solve the problem. The operation rules are stored in the rule base 31 and called as needed (step 17). EBL140 says "Why is it?"
Is known in the field of artificial intelligence as a method for deductively discovering general rules by explaining the inference engine 141 and the background knowledge 142. By using this method to explain why the operation plan currently under consideration works, we can obtain generalized operation rules. The present embodiment is characterized in that the generalized operation rule is linked to the case and stored (step 150), so that the operation rule to be used can be obtained quickly during the distribution planning. , Distribution plan 16
The same effect as that of the embodiment of FIG.

FIG. 10 shows an extended emergency network operating method using the emergency network operating method according to the present invention. First, a current situation analysis is performed to find out what kind of emergency situation is occurring (step 2) to grasp the situation. The analyzed current status is used to measure the similarity between the case and the current status when searching the cases stored in the case database 6 (step 4), and the most similar case is determined as the application case (step 8). ), The operation method belonging to the case and the current data are fused by computer processing to create an initial operation plan (step 10). The method for creating the initial operation plan is already known. Since the decision of the operation plan means that all the allocation problems are solved, the allocation plan can be virtually executed by the simulation 12. The result of the simulation 12 is evaluated (step 14), and if the plan satisfying the planner's request is prepared, the initial operation plan is adopted as it is. If the result of the simulation 12 is unsatisfactory, a correction method for correcting the case is selected (step 17). When the operator correction is selected, the operator corrects the case (step 15), and the correction result for the operation plan is returned to the simulation 12. When the distribution plan modification is selected, the cost is called (step 18), the input data is prepared (step 20), the distribution plan 16 is executed, and the result is returned to the simulation 12 as an operation modification plan. The present embodiment is characterized in that the distribution plan 16 can be selected as one of the correction methods in this scene. This makes it possible to always obtain a solution even for a distribution problem that is difficult by other methods, such as correction by the operator 15.

[0021]

According to the present invention, the speed and reliability of planning in an emergency operation plan of a transportation network connecting a plurality of supply points and consumption points is increased, and appropriate disaster initial countermeasures can be promptly taken. , Minimize damage and regulation to resource consumers.

[Brief description of drawings]

FIG. 1 is a diagram showing a processing flow of emergency operation planning according to an embodiment of the present invention.

FIG. 2 is a diagram showing details of a case database.

FIG. 3 is a diagram showing a processing flow for explaining a distribution plan.

FIG. 4 is a conceptual diagram of a multi-layered expansion network used for distribution planning.

FIG. 5 is a diagram illustrating a processing flow for explaining solution improvement processing.

FIG. 6 is a diagram showing a processing flow for explaining cost setting linked to a case in the preliminary examination.

FIG. 7 is a diagram showing a display example for setting a cost linked to a case in a preliminary examination.

FIG. 8 is a diagram showing a processing flow for explaining generalized operation rule creation linked to a case in a preliminary examination.

FIG. 9 is a display example of case contents.

FIG. 10 is a diagram showing a processing flow of an extended emergency network operation method as an application example of the present invention.

[Explanation of symbols]

 6 case database 22 case 24 case index 26 operation method 28 cost 30 cost database 52 source 54 sink 56-64 storage point 65-67 node 68-91 arc 21 case content display button 27 cost display button 29 cost function display Display 120 Preliminary study execution button 121 Target operation setting button 122 Cost setting button 123 Allocation plan execution button 124 Cost storage button 130 Cost setting editor

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Mariko Suzuki 52-1 Omika-cho, Hitachi-shi, Ibaraki Hitachi Ltd. Omika factory

Claims (6)

[Claims] 1. An operation plan for an emergency when a situation different from normal times occurs in a transportation network in which a plurality of supply points and consumption points are represented as nodes and a route for transporting a substance between nodes is represented as an arc. A case database that stores case examples that form a set of case index and operation method in case of emergency, which is obtained by pre-examination. A case corresponding to the case index that is the most similar to the current characteristic amount is searched, and a case corresponding to the case characteristic index is extracted, and a cost database that stores a cost for linking to the case operating method is linked to the extracted case operating method. Costs are extracted, and the extracted costs and various data necessary for planning are input, and an operation plan is created in the allocation plan. Emergency network operation wherein the Rukoto. 2. The emergency network operation method according to claim 1, wherein the allocation plan allocates a cost to each arc to obtain a minimum cost flow, and a target value that satisfies a realistic operational requirement item. The flow network correction method is performed according to the setting, and when the processing result does not satisfy the operation requirement item, the flow rate correction processing is performed again according to the target value change setting. 3. The emergency network operation method according to claim 1, wherein the preliminary examination executes an allocation plan according to the setting of operation request items and the setting of costs, and when the execution result is satisfactory, The case including the set operation request items and the set cost are stored in the case database and the cost database, respectively, and when the execution result is unsatisfactory, the cost setting or the operation request item setting is changed to make the distribution plan. An emergency network operation method characterized by executing. 4. The emergency network operation method according to claim 1, wherein when a case in the case database is designated, a case index and an operation method of the case, and a cost in a cost database linked to the operation method are displayed. An emergency network operation method characterized by displaying. 5. An operation plan for an emergency when a situation different from normal occurs in a transportation network in which a plurality of supply points and consumption points are expressed as nodes and a route for transporting a substance between nodes is expressed as an arc. A case database that stores case examples that form a set of case index and operation method in case of emergency, which is obtained by pre-examination. A rule database that stores the generalized operation rules generated based on the operating method and background knowledge of the case, which is searched based on the current characteristic amount, the case corresponding to the case index most similar to the current characteristic amount is extracted. From the above, an operation rule linked to the operation method of the extracted case is extracted, and the extracted operation rule is used as an operation plan for network operation. Emergency network operation wherein the Rukoto. 6. An operation plan planning in an emergency when a situation different from normal occurs in a transportation network in which a plurality of supply points and consumption points are represented as nodes and a route for transporting a substance between the nodes is represented as an arc. A case database that stores case examples that form a set of case index and operation method in case of emergency, which is obtained by pre-examination. The case is searched by the current feature quantity, the case corresponding to the case index most similar to the current feature quantity is extracted, the operation method of the extracted case and the current data are fused to create an initial operation plan, and the created The initial operation plan is evaluated by simulation, and if the result of the evaluation is not satisfied, the created initial operation plan is modified or the case operation method is used. The cost linked to the operating method of the extracted case is extracted from the cost database storing the cost linked to, the distribution plan is executed based on the extracted cost, the operation plan is prepared, and the modified initial operation plan or The network operation method for emergencies, characterized in that the planned operation plan is evaluated again by simulation, and if the result of the evaluation is satisfactory, the corrected initial operation plan or the planned operation plan is used as an operation plan for network operation. .