Support for Diagnosis: Spreadsheet View
From DSL
The Spreadsheet window is a special extension of GeNIe that is useful in rapid model building - all properties of every variable of a model are listed in one window and the user specifying a model can move rapidly between variables and enter or modify their specifications.
To open the Spreadsheet window, the user must click on the Spreadsheet icon 
located at the top right hand side of the taskbar in GeNIe.
The Spreadsheet window consists of several columns of information that describe the individual nodes (represented by rows) in detail. There are separate rows for each node and also for each state of the node. The information in the Spreadsheet View is also available in the Node Properties Sheet, However the Node Properties Sheet contains additional information that is not available in the Spreadsheet View, e.g., conditional probability tables.
The columns of the Spreadsheet are described below in the order from left to right. You can directly jump to the description of a particular column by clicking on the column in the image below.
Contents |
Node Name
Node Name is simply the name of the node that is being described. Each node, has a row for each state of the node. For example, say Battery', has two rows that corresponds to states: Pass and Fail. This column can be sorted alphabetically by clicking on the Node Name label. A node name can be changed manually by clicking on the box and typing the different name. In the figure above, the nodes are listed alphabetically starting with Battery and then Engine. These are the names given to two of the model nodes by the user.
State Name
The State Name column lists all states for each individual node. For example, node Battery has two states: Pass and Fail. For each of these two states, there are attached rows with information on each state relevant to the node. A state name can be changed manually by clicking on the box and by typing in a different name.
Target Name Format
Target name is an additional name assigned only for Target states of Target nodes. For Target Name Format, the user can choose between User Defined, Node Name, State Name, and Node+State Name and Inherit. If User Defined is chosen, then the user can type in a name of their choice. If Node Name is selected, then the node name will appear. If State Name is selected, then the Target state will appear. If Node+State Name is selected, then the Target node name will appear followed by the Target state name. This name may be chosen to be the same as a Target state name, as a Target node name, or as a Target node name followed by Target state name.
Target name format can be defined at a network level. This name is used for the node if we choose the option Inherit. The user can specify one of the options above at the network level in the General tab of Network properties Sheet in GeNIe. The selected format become then the default selection for all new nodes of the network.
See General tab section of Network properties Sheet for more information.
Special Name
Special Name column displays the special name for Target states of Target nodes. The name can be defined by using one of the Special Name Formats or can be specified by the user. When the user types anything in the box Special Name, the Special Name Format box for that node will be automatically changed to User Defined. If State Name were chosen, then only the state name would be used. Similarly, for Node Name, only the node name would be used. If the user were to choose Node Name + State Name, then the special name would be a concatenation of the two names. Special names can be changed manually by clicking on the box and by typing in a different name.
Node ID
The Node ID column gives a unique (across among all in the network) identifier for each node. Clicking on Node ID will change the order of the nodes from ascending to descending alphabetical order. Node IDs consist of a string of alphanumerical characters with no spaces (underscore characters are allowed).
State ID
The State ID is a unique identifier for each state of a particular node. The State ID for a particular state can be changed manually by clicking on the box and typing in a different name. State IDs consist of a string of alphanumerical characters with no spaces (underscore characters are allowed).
Prior Probability
Prior Probability displays the probabilities of the states of those nodes that are the root nodes of the network (i.e., nodes without parents). Many nodes representing faulty components fall into this category. Prior probabilities for components indicate how likely the components are to fail (it can be based on expert opinion or on objective data, such as average frequency of failure across the fleet). For example: a given component, A, may have a prior probability of being in the state Defective equal to 0.001 and of being in the state OK equal to 0.999. This prior probability means that only in one locomotive in a 1000 will the component A be defective per shop visit. To change the prior probabilities of a faulty component, locate the row of the node for which the probability is to be changed. Then locate the rows representing the states of the node. After finding the appropriate row, scroll right until you reach the column in the Spreadsheet labeled Prior Probabilities. Click on the box representing the probability of the state and type in the new probability. Enter the probabilities for the other states as well. The sum of prior probabilities for all states of a given node has to be equal to one. In the illustration above, the states, Starts, Overheats and Doesn't, for the node Engine have prior probabilities of 0.6, 0.1 and 0.3 respectively. If the prior probability of a node representing a faulty component is changed to zero then this fault is effectively eliminated from the process of diagnosis. In some cases, the faulty component nodes are dependent upon other nodes, such as nodes representing versions of the hardware, e.g., engine type. To change the probability of failure of such nodes, one needs to enter new values into the conditional probability table in the Definition Tab of Node Property Sheet.
See Definition Tab section of Node Property Sheet for more information on how to enter data into the Conditional Probability table.
Cost
The Cost column is defined for Observation nodes only. Cost contains a simple cost of a tests or an observation represented by the particular node. See the Cost of observation section in Support for Diagnosis for more information. The Observation cost can also be specified from the Node Properties Sheet for the particular node. See the Observation Cost section of Node Properties Sheet for more information on how to do this.
Type
Type describes the type of node. By clicking on Type, the rows can be sorted according to type. The type can be changed manually by clicking on the box and by typing in a different name. In the figure above, there are two types of nodes shown, Battery which is an Observation node and Engine which is a Target node.
You can also change a node type after creation either by right clicking on the node in the Graph View and selecting Change Type from the popup menu or selecting Change Type from the Node Menu in the Menu Bar. It will display a dialog box where you can choose the appropriate type for the node. See Change Type section of Node Menu for more information.
For more information on the functions of the various types of nodes, see Diagnostic Properties.
Ranked & Mandatory
The Ranked and Mandatory columns are used to represent subtypes of each node type. Checking and unchecking the boxes in corresponding columns specifies their subtype. If a box is checked in the Ranked column, then that node is ranked. If a box is checked in the Mandatory column, then that node is mandatory. The reverse applies for the boxes that are unchecked, or remain empty, i.e., they are not members of the Ranked or Mandatory subtypes. More than one subtype may be selected for a given node type, i.e., a node can be both Ranked and Mandatory. For example, in the figure above, the nodes Battery and Engine have both the subtype Ranked. However, if a box has been grayed out, then it does not pertain to the particular node type and cannot be changed. There are several type and subtype combinations that are legal and of interest to diagnosis: Target Ranked, Observation Ranked, Observation Set Default, and Observation Mandatory. These combinations are described in the Diagnostic Properties section.
The Ranked and Mandatory status can also be specified from the General tab in the Node Properties Sheet for the particular node. See the General Tab section of Node Properties Sheet for more information on how to do this.
Target State and Default State
For Target State and Default State, the user is able to select them by checking or unchecking the box. If the region is gray however, then that particular column does not pertain to the node. For example, the node, Battery is a Observation node hence the Target column is disabled [ grayed ]. For the Engine node, which is a Target node, the Target column is enabled and the Target state box is checked for both of the defective states of the node.
Default State, on the other hand, pertains to all nodes and by simply checking the box, it can be determined whether or not the node has been set to use the default state or not.
The Target and Default status can also be specified from the General tab in the Node Properties Sheet for the particular node. See the General Tab section of Node Properties Sheet for more information on how to do this.
Node Description and Question
The Node Description column contains a short text defining the node and its states. The Question column applies to Observation nodes only and describes in the form of a question that the observation is supposed to answer. In the Node Description column and the Question column, the user can click on the box labeled Edit or Add in each of the respective columns. Add indicates that there is no description or question available for the node, so the user can click on this tab to add new information to the column. Edit indicates that some text has been already defined for the node and can be updated or changed. By clicking on Edit or Add, a new window will open. It is in this window that the user can add their descriptions and questions for each particular node.
The Node Description and Question can also be specified from the Documentation tab in the Node Properties Sheet for the particular node. See the Documentation Tab section of Node Properties Sheet for more information on how to do this.
State Description and Treatment
The State Description column contains a short text defining the description of the state. The Treatment column describes how to treat the defect represented by the state. The Treatment column applies to Target states only and will be grayed out for all other states. The State Description and Treatment columns each have buttons that are labeled Edit or Add. By clicking on Add, the user can add new information regarding the state description or a Treatment. When clicking the Edit button in either of the columns, the user is able to add or alter descriptions of the nodes in the respective boxes.
The State Description and Treatment can also be specified from the Documentation tab in the Node Properties Sheet for the particular state. See the Documentation Tab section of Node Properties Sheet for more information on how to do this.
Node & State Links
The Link section has buttons labeled Edit or Add. These buttons can be utilized to edit the current link or to add a new link. These links in most cases are directed to a document with information that provides further documentation for the nodes and states. These documents can be used to repair procedures, schematics, block diagrams, manuals, etc. Links have names describing what the documents contain and the address of the document.
The Node and State Links can also be specified from the Documentation tab in the Node Properties Sheet for the particular node. See the Documentation Tab section of Node Properties Sheet for more information on how to do this.
